Daily reading queue

148 items for 2026-06-17 across 2 categories.

Active sources: 7. Sources represented in this queue: 6. The cron runs daily at 06:00 server time; arxiv RSS is often empty on weekends.

Linked to your results

score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.18220unread
Gatling: Rapid-Fire Consensus from Parallel Composition
Giulia Scaffino, Max Resnick, Joachim Neu · 2026-06-17
arXiv:2606. 18220v1 Announce Type: new Abstract: Consensus protocols form the core of blockchains and other replicated state machines, ensuring that all correct nodes process the same totally ordered log of input transactions.
Read next because Gatling: Rapid-Fire Consensus from Parallel Composition overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)". Matching terms: rect, under, correct, good, line, rate, without, chain. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.18220v1 Announce Type: new Abstract: Consensus protocols form the core of blockchains and other replicated state machines, ensuring that all correct nodes process the same totally ordered log of input transactions. In fault-free executions, performance is driven by the good-case transaction latency -- the time between a transaction becoming known to all nodes and its confirmation by the consensus protocol -- which depends on both how frequently proposals are made and, once made, how quickly they are confirmed. While prior work has established tight lower bounds on confirmation latency that modern protocols already achieve, it remains open whether the inter-proposal time can be further reduced below the state-of-the-art of one network delay. We introduce Gatling, an atomic broadcast protocol that achieves arbitrarily small inter-proposal times under rotating leader schedules; in particular, smaller than the network delay. Gatling runs multiple parallel instances of a black-box atomic broadcast protocol and staggers their proposal schedules to generate proposals in faster succession than state-of-the-art protocols. A deterministic interleaving rule merges the outputs of these instances into a single global log. We analyze the effects of head-of-line blocking caused by crashed leaders, and derive Gatling's optimal number of parallel instances. We further study the impact of Gatling on predictable validity and present two variants that retain this property. Finally, our experiments confirm that Gatling can be used with off-the-shelf component protocols to achieve low latency without fine-tuning the component protocol for minimum latency.
source ↗open
score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.18198unread
Seeing Is Not Screening: Multimodal Hidden Instruction Attacks on Agent Skill Scanners
Xiaojun Jia, Jie Liao, Simeng Qin, Ke Ma, Wenbo Guo, Yebo Feng, Aishan Liu, Yang Liu · 2026-06-17
arXiv:2606. 18198v1 Announce Type: new Abstract: Agent skills are emerging as an important attack surface in LLM-based systems.
Read next because Seeing Is Not Screening: Multimodal Hidden Instruction Attacks on Agent Skill Scanners overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, text, source, extraction, alone, does, chain, screen. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.18198v1 Announce Type: new Abstract: Agent skills are emerging as an important attack surface in LLM-based systems. Through an empirical study of existing skill scanners, we find that current defenses primarily rely on textual descriptions, manifests, and source code as the main signals for security analysis, which can leave visually conveyed malicious intent insufficiently examined. This creates a practical blind spot: harmful operational instructions hidden in images may bypass scanning while still being recoverable by multimodal agents during deployment. To systematically investigate this threat, we propose SkillCamo, a document-mediated multimodal instruction attack that conceals malicious instructions within images bundled with a skill while rewriting the surrounding documentation to naturally reference those images as part of the normal workflow. Thus, the attack does not rely on the image alone, but on the joint interpretation of textual guidance and visual payload at execution time. To defend against such attacks, we further propose ExecScan, an execution-grounded multimodal scanning module that performs intent extraction, behavior reconstruction, abuse assessment, and deliberative execution simulation over skill artifacts. ExecScan jointly analyzes documentation, code, referenced resources, and visual content to recover hidden instructions, reconstruct executable behavior chains, and identify downstream risks such as exfiltration, destruction, persistence, deception, and privilege escalation. Extensive experiments show that image-hidden malicious instructions challenge existing skill scanners, while ExecScan can improve the skill scanning performance.
source ↗open
score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.18120unread
Structural Role Injection in Handlebars-Templated LLM Prompts: Triple-Brace Interpolation, Delimiter Family, and the Limits of HTML Auto-Escaping
Mohammadreza Rashidi · 2026-06-17
arXiv:2606. 18120v1 Announce Type: new Abstract: Large language model applications build prompts from templates, and Handlebars is a widely used templating engine and the default prompt-template format in Microsoft Semantic Kernel.
Read next because Structural Role Injection in Handlebars-Templated LLM Prompts: Triple-Brace Interpolation, Delimiter Family, and the Limits of HTML Auto-Escaping overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)". Matching terms: latin, soft, assistant, rate, control, does, language, model. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.18120v1 Announce Type: new Abstract: Large language model applications build prompts from templates, and Handlebars is a widely used templating engine and the default prompt-template format in Microsoft Semantic Kernel. Its double-brace {{x}} expression HTML-escapes the interpolated value and is documented as the safe default; its triple-brace {{{x}}} expression inserts the value raw. We show that this choice silently governs an application's exposure to structural role injection, where attacker-controlled data carries chat role delimiters that forge a higher-privilege turn. A model-free analysis establishes the mechanism: Handlebars escaping rewrites angle brackets but not square brackets, colons, or Markdown hashes, so it neutralises ChatML, Llama-3, and XML role delimiters (survival rate 0.00) while leaving Llama-2 [INST], legacy Human:/Assistant:, and Markdown ### delimiters intact (survival rate 1.00 for the last two). We then run 5760 trials across seven delimiter families, two attack objectives, and four models (GPT-3.5 Turbo, GPT-4o mini, GPT-4.1 mini, Claude Haiku 4.5) at a combined API cost of 1.63 USD. GPT-3.5 Turbo follows the task-hijack instruction in 97% of raw and 91% of escaped trials, with the escaping protection concentrated in the angle-bracket families and absent for the colon- and Markdown-based families; the harder secret-exfiltration objective, which does not saturate, exposes the same family interaction more cleanly. Claude Haiku 4.5 resists both objectives almost entirely. The escaped default protects only the delimiter schemes whose characters HTML escaping happens to cover, gives no protection for the rest, and cannot substitute for a structural separation of instruction and data.
source ↗open
score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.17957unread
Children Are Not the Enemy: Child-Fit Security as an Alternative to Bans and Surveillance
Kopo M. Ramokapane, Rui Huan, Zaina Dkaidek, Awais Rashid · 2026-06-17
arXiv:2606. 17957v1 Announce Type: new Abstract: Digital technologies are now central to children's learning, play, communication, identity formation, and social participation.
Read next because Children Are Not the Enemy: Child-Fit Security as an Alternative to Bans and Surveillance overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, line, control, screen, contexts. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.17957v1 Announce Type: new Abstract: Digital technologies are now central to children's learning, play, communication, identity formation, and social participation. Yet dominant approaches to children's online safety often rely on containment mechanisms, including bans, age gates, parental controls, monitoring, and screen-time restrictions. These approaches can be useful in specific contexts, but they often frame child protection primarily as a problem of restricting access to systems designed for adults. In this paper, we argue that this framing is inadequate for children's digital lives and insufficient as a security paradigm. We propose Child-fit security, a design paradigm in which technologies likely to be used by children treat a child as legitimate users, not attackers to be excluded, vulnerabilities to be patched, or risks to be managed. In this paradigm, children's wellbeing, development, privacy, safety, agency, and rights become core security requirements. This shifts the focus of protection from apps, accounts, and data to the child-system relationship, which means protecting both the child and their participation. We conceptualise child-fit security, contrast it with containment-oriented approaches, define its core principles, and discuss its implications for security design. We conclude by presenting a research agenda for making child-fit security operational.
source ↗open
score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.17555unread
An AI Security Agent for Banking: Multi-Vector Fraud and AML Detection Across Retail and Corporate Accounts
Joseph Walusimbi, Joshua Benjamin Ssentongo · 2026-06-17
arXiv:2606. 17555v1 Announce Type: new Abstract: Banks simultaneously face signature-based fraud (card-not-present attacks, account takeover, ATM cloning) and behavioural financial crime (structuring, layering, mule networks, business email compromise) -- two threat families with fundamentally different detection requirements.
Read next because An AI Security Agent for Banking: Multi-Vector Fraud and AML Detection Across Retail and Corporate Accounts overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: rect, under, assistant, line, rate, model. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.17555v1 Announce Type: new Abstract: Banks simultaneously face signature-based fraud (card-not-present attacks, account takeover, ATM cloning) and behavioural financial crime (structuring, layering, mule networks, business email compromise) -- two threat families with fundamentally different detection requirements. Static rule engines that reliably catch brute-force and high-velocity events are structurally blind to business-email-compromise (BEC) payment redirection, session hijacking, and money-laundering layering, which are engineered to appear indistinguishable from legitimate activity at the individual transaction or session level. This paper presents an AI security agent for retail and corporate banking that addresses this gap through a three-component fusion architecture operating on two parallel event streams: a transaction stream (card fraud, ACH/wire fraud, AML categories) and a session stream (account takeover, session hijacking, SIM-swap, insider abuse). Each stream combines an LSTM sequence model capturing per-account behavioural history, a statistical velocity/threshold monitor, and a graph/network module capturing account-counterparty relationship patterns (fan-in, fan-out, pass-through ratio) for money-laundering detection. Experiments on a synthetic event log of 237,669 transactions and 113,508 sessions across 13 threat categories and 3,470 simulated accounts demonstrate overall F1 of 0.787 (transaction stream) and 0.867 (session stream) for the proposed model, versus 0.562/0.733 for a rule-based baseline and 0.655/0.713 for an LSTM-only baseline. The agent includes a customer-facing transaction-verification chatbot (96.6% identity verification accuracy, 86.8% mass-reset attack detection) and an analyst case-summary assistant (99.3% action-recommendation F1), with Critical-tier automated response latency under 0.43 ms at the 95th percentile.
source ↗open
score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.17421unread
Bifrost: Hybrid TEE-FHE Inference for Privacy-Preserving Transformer and LLM Serving
Chenghao Chen, Kailun Qin, Xiaolin Zhang, Chi Zhang, Dawu Gu · 2026-06-17
arXiv:2606. 17421v1 Announce Type: new Abstract: Cloud-hosted transformer and large language model (LLM) inference creates a direct confidentiality problem: user prompts may contain sensitive code, business data, personal information, or regulated documents, yet remote serving exposes intermediate state to the cloud software stack and accelerator runtime.
Read next because Bifrost: Hybrid TEE-FHE Inference for Privacy-Preserving Transformer and LLM Serving overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, text, fill, persona, rect, soft, line, project. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.17421v1 Announce Type: new Abstract: Cloud-hosted transformer and large language model (LLM) inference creates a direct confidentiality problem: user prompts may contain sensitive code, business data, personal information, or regulated documents, yet remote serving exposes intermediate state to the cloud software stack and accelerator runtime. Fully homomorphic encryption (FHE) keeps accelerator-side execution ciphertext-only, but end-to-end LLM inference remains expensive because linear layers are interleaved with non-linear, cache-state, and refresh-sensitive operators. CPU trusted execution environments (TEEs) can execute those operators natively, but a CPU TEE alone does not define how an untrusted accelerator should participate. We present Bifrost, a hybrid TEE-FHE serving architecture in which secrets are provisioned only to an attested CPU TEE, while the accelerator, device memory, driver/runtime stack, and host software remain outside the trusted computing base. Bifrost uses FHE as a secure delegation mechanism for projection and feed-forward linear layers on accelerator-backed CKKS, while non-linear operators, attention-side control logic, KV-state transitions, and decrypt-then-encrypt refresh execute inside the CPU TEE. Bifrost+ further applies a prefill/decode split: prompt-side KV state is built inside the CPU TEE, and only decode-side state enters the hybrid ciphertext path. In an estimator-style comparison matching Euston's methodology, Bifrost reduces projected latency by 9.25x on GPT-2 (1.5B) and 9.91x on LLaMA 3 (8B). In direct CKKS/FHE deployments, Bifrost+ reduces TTFT by 14.6-45.8x on GPT-2 (124M) and 15.3-53.4x on Qwen3 (0.6B). The systems lesson is selective encrypted execution: use FHE only where ciphertext-only accelerator delegation is required, and keep non-linear, refresh, and prompt-side work inside the CPU TEE.
source ↗open
score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.17358unread
OTRO: Oblivious Tokenization Path with Square-Root ORAM
Jonghyun Lee, Yongqin Wang, Rachit Rajat, Daniel Wong, Mengyuan Li, Murali Annavaram · 2026-06-17
arXiv:2606. 17358v1 Announce Type: new Abstract: The CPU-side large language model (LLM) tokenizer is a critical security gap in LLM serving through a confidential computing stack with CPU and GPU trusted execution environments (TEEs).
Read next because OTRO: Oblivious Tokenization Path with Square-Root ORAM overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)". Matching terms: fill, under, source, token, rate, implement, leakage, language. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.17358v1 Announce Type: new Abstract: The CPU-side large language model (LLM) tokenizer is a critical security gap in LLM serving through a confidential computing stack with CPU and GPU trusted execution environments (TEEs). Tokenizers converts the prompts through table-driven lookups, and the resulting memory access patterns are a powerful source of side-channel leakage. Recent work demonstrates end-to-end recovery of user prompts from tokenizer access pattern on production Intel TDX. However, a drop-in use of the popular tree-based Oblivious RAMs (e.g., PathORAM) to prevent access-pattern leakage introduces $\sim$13$\times$ tokenizer slowdown, resulting in 10-58% higher time-to-first-token (TTFT). In this paper, we present OTRO, an efficient, oblivious tokenization path tailored to latency-critical LLM serving. OTRO relies on square-root ORAM for fast single-access lookups, but avoids its prohibitive $O(N\log^2N$) rebuild cost every $\sqrt{N}$ accesses through three key innovations. First, OTRO provides a pool of replicated square-root ORAM instances that utilize the read-only nature of tokenizer table. Second, an epoch-based rotation policy decouples accesses from rebuilds and pads each epoch with dummy accesses to its boundaries, minimizing observable information. Lastly, chunked KV-cache-aware tokenization further overlaps rebuilds with GPU prefill and minimizes the instance count. Implemented as modules in HuggingFace Tokenizers and nano-vLLM, running within a TDX-enabled CVM with an NVIDIA H100 GPU, OTRO limits TTFT overhead to at most 4.5%, keeps tokenizer-induced latency under 10\% of total TTFT, and adds less than 0.5 GB of memory overhead while reducing the tokenizer's observable leakage across various model families and sizes.
source ↗open
score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.17110unread
Loss Landscape Poisoning: Targeted Extraction of Unseen Training Data from LLMs
Md Abdullah Al Mamun, Ngoc Phu Doan, Pedram Zaree, Ihsen Alouani, Nael Abu-Ghazaleh · 2026-06-17
arXiv:2606. 17110v1 Announce Type: new Abstract: Large Language Models are increasingly trained on proprietary or sensitive data, from private healthcare and financial records to user conversations containing secrets.
Read next because Loss Landscape Poisoning: Targeted Extraction of Unseen Training Data from LLMs overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: rect, rate, extraction, trained, leakage, completion, language, model. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.17110v1 Announce Type: new Abstract: Large Language Models are increasingly trained on proprietary or sensitive data, from private healthcare and financial records to user conversations containing secrets. Ensuring the privacy of such data against extraction attacks has become a central concern. In this paper, we ask whether an attacker who can poison a portion of the training data can facilitate the leakage of a separate target record they have no access to. We answer in the affirmative and show that such leakage can be induced by a poisoning mechanism that reshapes the model's local loss landscape around the target completion. Our key insight is that poisoning to create a sharp loss minimum at the target, surrounded by elevated loss on nearby alternatives, forces the model to memorize the target as the unique low-loss solution in its neighborhood. The attack requires no architectural changes, and generalizes across centralized and federated learning settings. We demonstrate that the attack amplifies privacy leakage across language (up to 100% successful extraction), and vision-language models (up 90% successful extraction). We show that the attack is thwarted when the model is trained to be differentially private. However, we introduce a new attack that directly probes the loss landscape bypassing even differential privacy defenses.
source ↗open
score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.17089unread
Security and Human-Centered Assessment of BACnet-Controlled DALI Infrastructure in an Educational Building Automation Testbed
Ariton Verush · 2026-06-17
arXiv:2606. 17089v1 Announce Type: new Abstract: Building automation and control systems integrate heating, ventilation, air conditioning, lighting, sensing, and management functions through specialized communication protocols.
Read next because Security and Human-Centered Assessment of BACnet-Controlled DALI Infrastructure in an Educational Building Automation Testbed overlaps with clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "Add C2 control arm (donor sees marker_B without marker_A) to disambiguate paired-marker binding from marker_B leaking alone", experiment "#351 follow-up: broader-vocab position-0 sweep at T=1.0 + position-1 suffix isolation". Matching terms: rate, control, test, lora, model. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.17089v1 Announce Type: new Abstract: Building automation and control systems integrate heating, ventilation, air conditioning, lighting, sensing, and management functions through specialized communication protocols. While this integration enables flexible building operation, it also creates complex cyber-physical environments that are difficult to inspect, secure, and explain to new analysts. This paper presents a practical security and human-centered case study of a BACnet/IP building automation testbed with DALI lighting infrastructure, investigated during a domotics-oriented cybersecurity hackathon in Thun, Switzerland in April 2026. The study combines network-oriented enumeration, object-level inspection, physical rack analysis, and reflective HCI analysis of tool-supported learning. Using Yabe and BACteria, the work documents observable BACnet services, reconstructs structured object hierarchies, identifies room-level lighting-control paths, and maps BACnet objects to DALI group-level infrastructure. The analysis emphasizes that BACS assessment is not only a technical protocol task: it also requires usable tool interfaces, physical observability, interpretable naming conventions, and safe mental models for command priorities. The paper contributes a compact case study of BACnet/DALI exploration in an educational testbed and discusses implications for cybersecurity education, human-centered security tooling, and responsible experimentation in cyber-physical building environments.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17696unread
FllumaOne: A Code-Native Multimodal CAD Dataset with Executable Programs and Kernel-Validated Feature Histories
Jizong Zhan · 2026-06-18
arXiv:2606. 17696v1 Announce Type: new Abstract: Parametric computer-aided design records both final geometry and the ordered construction history that determines how a part can be edited.
Read next because FllumaOne: A Code-Native Multimodal CAD Dataset with Executable Programs and Kernel-Validated Feature Histories overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, eval, line, rate, trained, test, lora, qwen2. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17696v1 Announce Type: new Abstract: Parametric computer-aided design records both final geometry and the ordered construction history that determines how a part can be edited. Datasets for editable CAD research should therefore expose modeling operations, parameters, and feature dependencies together with validated geometry. We introduce FllumaOne, a code-native multimodal CAD dataset whose models are generated by executable Python programs in Flluma, a Qt/C++ OpenCASCADE-based CAD system. Each sample aligns its program with a structured feature tree, a training-oriented intermediate representation, STEP geometry, a surface point cloud, natural-language descriptions, metadata, and eight canonical visible-edge renderings. The primary release, FllumaOne-100K, contains 100,000 accepted samples across four template-level complexity regimes. Programs are executed and retained only after kernel geometry, solid validity, and export checks; release reports also record modality completeness and split-level duplicate tests. A Qwen2.5-Coder-1.5B LoRA baseline trained on 80,000 samples achieves 99.98% Python syntax validity, 99.97% Flluma build success, and 99.14% STEP-export validity on the held-out 10,000-sample test split. For the 9,909 predictions converted to surface point clouds, the mean normalized Chamfer Distance is 0.002124. The dataset supports conditioned CAD reconstruction, executable program synthesis, feature-tree prediction, B-Rep analysis, retrieval, design completion, and editable reverse engineering.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17637unread
Brick-DICL: Dynamic In-Context Learning for Automated Brick Schema Classification
Yiyue Qian, Shinan Zhang, Huan Song, Negin Sokhandan, Hannah Marlowe, Diego Socolinsky · 2026-06-18
arXiv:2606. 17637v1 Announce Type: new Abstract: Building Management Systems (BMS) are essential for optimizing energy efficiency and operational performance in modern buildings.
Read next because Brick-DICL: Dynamic In-Context Learning for Automated Brick Schema Classification overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, class, rate, implement, compare, stage, test, language. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17637v1 Announce Type: new Abstract: Building Management Systems (BMS) are essential for optimizing energy efficiency and operational performance in modern buildings. However, the lack of standardization across BMS points from different manufacturers creates significant barriers to integration and data utilization. While the Brick schema offers a standardized ontology for building systems, mapping BMS points to appropriate Brick classes presents three critical challenges: (i) the extensive number of Brick classes (936 in the latest version), (ii) limited domain-specific knowledge in large language models (LLMs), and (iii) substantial manual effort required for verification. To address these challenges, we propose Brick-DICL, a two-stage dynamic in-context learning framework for automated Brick schema classification. Brick-DICL consists of two primary components: metadata-RAG, which retrieves relevant examples to enhance LLMs' domain knowledge, and class-RAG, which narrows down potential Brick classes to address the large classification space. Additionally, we implement a multi-LLM filtering mechanism that compares predictions across multiple models, flagging low-confidence classifications for human review. As a result: (i) General: Brick-DICL is applicable to any building management system regardless of manufacturer or metadata format; (ii) Novel and Powerful: as the first dynamic in-context learning approach for Brick schema classification, Brick-DICL achieves significant classification accuracy improvements on building datasets, outperforming existing methods; (iii) Efficient: our multi-LLM filtering strategy reduces manual verification effort, enabling rapid digital building onboarding. Extensive experiments demonstrate Brick-DICL's effectiveness across diverse building datasets, accelerating the path toward standardized, interoperable building management systems.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17405unread
Treatment Response Optimized Clinical Decision Support AI System via Digital Twin Simulation
Xinyu Qin, Anil K. Sood, Ruiheng Yu, Sara Corvigno, Elaine Stur, Lu Wang · 2026-06-18
arXiv:2606. 17405v1 Announce Type: new Abstract: Clinical decision support AI systems (CDSASs) must adapt to evolving patient conditions in real-time while adhering to strict safety constraints.
Read next because Treatment Response Optimized Clinical Decision Support AI System via Digital Twin Simulation overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, persona, line, rate, compare, trained, model. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17405v1 Announce Type: new Abstract: Clinical decision support AI systems (CDSASs) must adapt to evolving patient conditions in real-time while adhering to strict safety constraints. We present an online adaptive framework that integrates Treatment Effect (TE) estimation to quantify clinical benefits, a patient Digital Twin (DT) to simulate treatment trajectories, and Reinforcement Learning (RL) for sequential decision-making. The AI system is initially trained on historical medical records and operates in a continuous learning loop. To ensure safety, a rule-based module monitors vital signs and blocks contraindicated treatments. Cases with strong internal model disagreement are flagged for clinician review, simulated in our experiments via a pre-trained outcome model. We validate our framework using both a synthetic clinical simulator and a real-world ovarian cancer dataset from The Cancer Genome Atlas (TCGA). In both simulated and clinical settings, our method demonstrated superior effectiveness and stability in recommending treatments compared to standard computational baselines. Furthermore, the AI system maintains low latency and requires expert consultation for only a minority of cases in our experimental validation, demonstrating its potential as a safe, clinician-supervised tool for personalized medicine that continuously improves through practical use.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17368unread
Distributed General-Purpose Agent Networks: Architecture, Key Mechanisms, and Prototypes
Shengli Zhang, Deen Ma, Zibin Lin, Taotao Wang · 2026-06-18
arXiv:2606. 17368v1 Announce Type: new Abstract: Large language models have accelerated the transition from passive conversational assistants to autonomous agents that can understand goals, plan actions, invoke tools, and execute multi-step tasks.
Read next because Distributed General-Purpose Agent Networks: Architecture, Key Mechanisms, and Prototypes overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)". Matching terms: persona, under, assistant, rate, binding, propagate, trained, capability. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17368v1 Announce Type: new Abstract: Large language models have accelerated the transition from passive conversational assistants to autonomous agents that can understand goals, plan actions, invoke tools, and execute multi-step tasks. Yet the capability of a single agent remains constrained by its local data, tool permissions, runtime environment, and governance boundary. This paper studies distributed general-purpose agent networks: open peer-to-peer networks in which heterogeneous agents deployed on personal devices, edge nodes, or autonomous computing environments can discover one another, establish trust, negotiate cooperation rules, and execute open-ended tasks. We argue that such networks cannot be obtained by simply combining existing peer-to-peer overlays with conventional multi-agent systems. Unlike traditional P2P networks, agent networks must propagate semantic declarations about intentions, capabilities, states, and cooperation constraints. We therefore propose a layered architecture centered on a protocol adaptation layer that connects upper-level task semantics with lower-level network operations. Based on this architecture, the paper identifies three core mechanism problems: semantic announcement propagation for collaborator discovery, verifiable identity and multi-topic reputation for cooperation governance, and semantic-gradient mechanism design for open task execution. For each problem, we present a technical route, including bodyless gossip with sequential logs, BAID-based identity binding with MG-EigenTrust reputation, and a Stackelberg-style mechanism-generation loop driven by semantic attribution feedback. We further report prototype overhead results for BAID-style tiered verification and mechanism-level simulations of MG-EigenTrust under cross-topic disguise-collusion attacks. The resulting framework provides a system-level foundation for open, trustworthy, and scalable agent collaboration.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17289unread
Nothing from Something: Can a Language Model Discover 0?
Phoebe Zeng, Thomas L. Griffiths, Brenden M. Lake · 2026-06-18
arXiv:2606. 17289v1 Announce Type: new Abstract: AI systems based on artificial neural networks are being developed with aspirations of pushing the boundary of human mathematical knowledge.
Read next because Nothing from Something: Can a Language Model Discover 0? overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, eval, test, language, model. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17289v1 Announce Type: new Abstract: AI systems based on artificial neural networks are being developed with aspirations of pushing the boundary of human mathematical knowledge. A key question for these systems is how much they can reach beyond their training data. Mathematical discovery requires a strong form of out of distribution generalization; the ability to hypothesize genuinely new - and potentially logically more powerful - mathematical structures. It has been hypothesized that language abilities support such generalizations in human cognition. In this work, we use simple arithmetic as a case study for examining how modern AI models could expand their mathematical horizons, evaluating whether these models can independently discover the concept of "zero". We show that We show that (1) language models of a GPT-2 size are unable to perform this generalization at test time regardless of language pretraining, but (2) models can improve substantially after training on tens or hundreds of examples of zero. Additionally, we find that language pretraining reduces the number of required examples by approximately $50\%$, showing that language abilities can scaffold mathematical discovery in neural models.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15378unread
Rethinking the Role of Efficient Attention in Hybrid Architectures
Ziqing Qiao, Yinuo Xu, Chaojun Xiao, Zhou Su, Zihan Zhou, Yingfa Chen, Xiaoyue Xu, Xu Han, Zhiyuan Liu · 2026-06-17
arXiv:2606. 15378v1 Announce Type: new Abstract: Modern language models increasingly adopt hybrid architectures that combine full attention with efficient attention modules, such as sliding-window attention (SWA) and recurrent sequence mixers.
Read next because Rethinking the Role of Efficient Attention in Hybrid Architectures overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, under, eval, full, capability, language, model. Source: arxiv cs.CL (NLP).
arXiv:2606.15378v1 Announce Type: new Abstract: Modern language models increasingly adopt hybrid architectures that combine full attention with efficient attention modules, such as sliding-window attention (SWA) and recurrent sequence mixers. However, how these efficient modules shape model capabilities remains poorly understood. To address this gap, we conduct a systematic analysis across hybrid architectures from three perspectives: scaling behavior, mechanism analysis, and architecture design. First, from a scaling perspective, we find that efficient-attention design primarily affects how fast long-context capability emerges, while different hybrids eventually converge to comparable long-context performance under sufficient training. Second, mechanistically, we show that long-range retrieval is mainly carried by full attention, whereas efficient attention shapes its optimization trajectory. This explains a counter-intuitive phenomenon we call Large-Window Laziness: larger SWA windows can delay the formation of retrieval heads in full-attention layers. Third, guided by this mechanism, we show that applying NoPE to only the full-attention layers of a small-window SWA hybrid substantially improves long-context performance with negligible impact on short-context performance.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15325unread
Prior over Evidence: Stereotype-Driven Diagnosis in LLM-Based L2 Pronunciation Feedback
Rong Wang, Kun Sun · 2026-06-17
arXiv:2606. 15325v1 Announce Type: new Abstract: Large language models are increasingly deployed for written pronunciation feedback in second-language (L2) English learning, under the assumption that their diagnoses are grounded in the supplied speech evidence rather than in priors from pretraining.
Read next because Prior over Evidence: Stereotype-Driven Diagnosis in LLM-Based L2 Pronunciation Feedback overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, rect, under, alignment, correct, wrong, eval, line. Source: arxiv cs.CL (NLP).
arXiv:2606.15325v1 Announce Type: new Abstract: Large language models are increasingly deployed for written pronunciation feedback in second-language (L2) English learning, under the assumption that their diagnoses are grounded in the supplied speech evidence rather than in priors from pretraining. This assumption is tested on 1,800 L2-Arctic utterances spanning six L1 backgrounds, three audio-capable LLMs, four pronunciation dimensions, and five evidence conditions ranging from a text-only baseline to numeric acoustic features and raw audio. Each (utterance x model x condition x dimension) cell is scored on three metrics: Rating Accuracy (RA) against gold labels, Evidence Coherence (EC) assessing internal consistency without ground truth, and Grounded Correctness (GC) evaluated against gold evidence. Results show three findings across models. First, rating accuracy and grounded reasoning decouple: 39.6% of judged cells contain internally coherent reasoning that supports a wrong rating, against only 15.8% where the reasoning supports a correct rating. Second, phoneme-level feedback converges to a fixed inventory of L2-English difficulty phones that recurs across all six L1 backgrounds and all evidence conditions. Third, acoustic evidence improves the rating only when the supplied feature directly probes the target dimension: textualised F0 range raises pitch-variation grounding from (0.18-0.19) to (0.45-0.62) across all three models, while stress and phoneme correctness, which require target-to-realisation alignment, remain ungrounded. The same audio waveform without textualised F0 values does not reproduce this improvement. These findings indicate that current general-purpose LLMs are more reliable as verbalisers of externally computed pronunciation evidence than as standalone diagnostic engines.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15216unread
Spokes: Optimizing for Diverse Pretraining Data Selection
Clarence Lee, Yejin Choi, Luke Zettlemoyer, Pang Wei Koh, Hai Leong Chieu · 2026-06-17
arXiv:2606. 15216v1 Announce Type: new Abstract: Diversity plays a critical role in data selection, improving performance under fixed data budgets by reducing redundancy and repetition.
Read next because Spokes: Optimizing for Diverse Pretraining Data Selection overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, rect, under, eval, line. Source: arxiv cs.CL (NLP).
arXiv:2606.15216v1 Announce Type: new Abstract: Diversity plays a critical role in data selection, improving performance under fixed data budgets by reducing redundancy and repetition. However, optimizing for diversity is inherently challenging, as it is a set-level property that depends on interactions between data points rather than individual examples. As a result, existing approaches typically rely on proxies or approximations, which often fail to ensure sufficiently diverse subsets. In this work, we directly optimize diversity by introducing a probabilistic diversification framework based on the G-Vendi score, optimized via exponentiated gradient descent. Our method produces subsets that are substantially more diverse than those obtained via random sampling, achieving a +489 increase in G-Vendi score on a 500k-sample subset. We evaluate our approach on FineWeb and DCLM, where it consistently outperforms existing methods. Notably, SPOKES (diversity-only) improves average downstream performance by +0.4 and +0.5 points over random sampling on DCLM and FineWeb, respectively. More importantly, jointly optimizing for both quality and diversity yields the strongest results: SPOKES achieves gains of +1.5 and +1.4 points on DCLM and FineWeb, outperforming all baselines, including semantic deduplication and quality filtering.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15161unread
Beyond Layer Importance in Layer-wise Sparsity: An Inter-Layer Perturbation-Absorption Perspective
Tao Jing, Ningxin Wu, Chen Kang, Dong Yu, Changliang Li, Pengyuan Liu · 2026-06-17
arXiv:2606. 15161v1 Announce Type: new Abstract: The considerable layer-wise redundancy in large language models (LLMs) has established non-uniform sparsity allocation across layers as the standard pruning approach for efficient compression.
Read next because Beyond Layer Importance in Layer-wise Sparsity: An Inter-Layer Perturbation-Absorption Perspective overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: rect, under, correct, alpha, middle, line, rate, control. Source: arxiv cs.CL (NLP).
arXiv:2606.15161v1 Announce Type: new Abstract: The considerable layer-wise redundancy in large language models (LLMs) has established non-uniform sparsity allocation across layers as the standard pruning approach for efficient compression. Existing layer-wise allocation methods that estimate allocation strategy from local signals such as activation outliers or weight spectra mainly derive from local layer importance, whereas the final post-pruning performance is also influenced by the network's subsequent compensatory capacity. In this paper, we directly characterize this property through controlled perturbation experiments. We make the following empirical findings. First, layers exhibit highly heterogeneous responses to pruning-scale perturbations. In most cases, early layers amplify perturbations, while middle and late layers actively absorb them, with relative L2 drift decreasing monotonically across depth and direction realigning toward the unperturbed hidden-state trajectory. Second, absorption is a large-perturbation phenomenon. Under small perturbations the network exhibits amplification across all layers, and the transition to absorption occurs smoothly as perturbation magnitude grows to pruning scale. This enriches the linearized accumulation theory underlying related works. Building on these findings, we define an absorption coefficient per layer and propose absorption-aware correction, an orthogonal augmentation that improves OWL and AlphaPruning by reducing perplexity by 7.13% and boosting zero-shot accuracy by 1.02% across multiple model families at 70% sparsity.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15079unread
Ling and Ring 2.6 Technical Report: Efficient and Instant Agentic Intelligence at Trillion-Parameter Scale
Ang Li, Ben Liu, Bin Han, Bin Hu, Bin Jing, Binbin Hu, Bing Li, Cai Chen, Caizhi Tang, Changxin Tian, Chao Huang, Chao Zhang, Chen Liang, Chen Qian, Chengfu Tang, Chengyao Wen, Chilin Fu, Chunwei Wu, Cong Zhang, Cunyin Peng, Daixin Wang, Dalong Zhang, Deng Zhao, Dingnan Jin, Dingyuan Zhu, Donghao Zhang, Fan Yuan, Fangzheng Zhao, Fanzhuang Meng, Feifan Wu, Feng Xu, Fengbin Fang, Gangshan Wang, Guodong Yang, Hailin Zhao, Haitao Wang, Haitao Zhang, Hanxiao Zhang, Hanzi Wang, Hao Dai, Hao Liu, Hao Qian, Hao Wu, Haoxiong Liu, Haoyu Xu, Heng Zhang, Hong Liu, Hongliang Zhang, Hongrui Liu, Hongxun Li, Hongzhi Ruan, Huaidong Xiong, Huihuang Zheng, Huikang Tang, Jia Guo, Jia Li, Jia Liu, Jiameng Wang, Jiaming Liu, Jiannan Shi, Jianping Wei, Jiaolong Yang, Jiapeng Wang, Jie Gao, Jie Wang, Jiewei Wu, Jin Yang, Jinjin Li, Jinjing Huang, Jinquan Sun, Jinyao Chen, Juanhui Tu, Jun Liu, Jun Mei, Jun Xu, Jun Zhou, Junjie Ou, Junnan Sipan, Junpeng Fang, Kaihong Zhang, Kaiqin Hu, Ke Shi, Kuan Xu, Kun Tang, Kunlong Chen, Lanyin Mei, Lei Chen, Lei Liang, Lei Xu, Li Tang, Liang Jiang, Liangcheng Fu, Lihui Zhang, Linfeng Shi, Lintao Ma, Liyuan Liu, Longfei Li, Longfei Zheng, Lu Liu, Lu Yu, Man Li, Meiqi Zhu, Meng Li, Mengjie Gao, Mengshu Sun, Mingming Yin, Mingyang Zhang, Mingyuan Fan, Nuo Xu, Pan Tang, Peijie Jiang, Peilong Zhao, Peng Lin, Pingping Liu, Qi Zuo, Qian Zhao, Qiang Cheng, Qianggang Cao, Qiaoben Bao, Qing Cui, Qingyuan Yang, Qitao Shi, Qiyin Huang, Qizheng Zhou, Quan Wan, Runyuan Zhao, Shaomian Zheng, Shaowei Wei, Shengnan Zhang, Shuaicheng Li, Shujie Li, Shuo Zhang, Sikang Bian, Tianchu Yao, Tiange Xu, Tianshu Wang, Ting Guo, Tinghao Wang, Tingwei Huang, Tong Zhao, Tongkai Yang, Wang Hong, Wanli Gu, Wei Lu, Weichang Wu, Weiguang Han, Weiquan Li, Wenbo Shen, Wenjing Fang, Wenzhi Tang, Xiang Shu, Xiao Shi, Xiaodong Yan, Xiaolu Zhang, Xiaopei Wan, Xiaqing Sun, Xin Zhao, Xingyu Lu, Xinxing Yang, Xinyao Tang, Xinyu Kong, Xinyu Liu, Xiong Xu, Xuan Sun, Xudong Han, Xudong Wang, Xujie Shen, Yalin Zhang, Yangyang Hou, Yankun Ren, Yao Zhao, Ye Chen, Yeyang Chen, Yibo Cao, Yifan Zuo, Yijie Chen, Ying Li, Yingjie Song, Yingxue Li, Yiqi Wang, Yixuan Sun, Yizhu Xiao, Yongfei Xu, Yu Liu, Yuchen Fang, Yue Gao, Yue Yu, Yue Zhang, Yuqi Zhang, Yuxiao He, Yuxiao Lu, Yuxin Tian, Yuxuan Li, Yuzhuo Fu, Zhankai Xu, Zhaoxin Huan, Zhenduo Zhang, Zhengke Gui, Zhengyu Huang, Zhenjun Ma, Zhenxuan Pan, Zheping Qu, Zhibo Zhu, Zhidong Fan, Zhigang Huangfu, Zhihao Wang, Zhiqiang Zhang, Zhizhen Liu, Zhuyan Zhou, Zibin Lin, Zihang Zeng, Zihao Wang, Zilong Wang, Ziqi Liu, Zitao Xuan, Zixuan Cheng, Zujie Wen, Zuoli Tang · 2026-06-17
arXiv:2606. 15079v1 Announce Type: new Abstract: Efficient and scalable agentic intelligence requires models that can deliver both low-latency responses and strong reasoning capabilities while remaining practical to train, serve, and deploy.
Read next because Ling and Ring 2.6 Technical Report: Efficient and Instant Agentic Intelligence at Trillion-Parameter Scale overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, text, rect, alignment, correct, source, token, line. Source: arxiv cs.CL (NLP).
arXiv:2606.15079v1 Announce Type: new Abstract: Efficient and scalable agentic intelligence requires models that can deliver both low-latency responses and strong reasoning capabilities while remaining practical to train, serve, and deploy. In this report, we present Ling-2.6 and Ring-2.6, a family of models designed to address this challenge at scale. Ling-2.6 is optimized for instant response generation and high capability per output token, whereas Ring-2.6 is tailored for deeper reasoning and more advanced agentic workflows. Instead of training from scratch, we upgrade the Ling-2.0 base model through architectural migration pre-training and large-scale post-training. This upgrade is guided by a unified co-design of model architecture, optimization objectives, serving systems, and agent training environments, enabling improvements in both model capability and deployment efficiency. At the architectural level, we introduce a hybrid linear attention design that integrates Lightning Attention with MLA, improving the efficiency of long-context training and decoding. To further enhance token efficiency, we optimize capability per output token through Evolutionary Chain-of-Thought, Linguistic Unit Policy Optimization, bidirectional preference alignment, and shortest-correct-response distillation. For agentic capabilities, we propose KPop, a reinforcement learning framework designed to support stable training of Ring-2.6-1T on large-scale environment-grounded data. KPop improves training efficiency through asynchronous scheduling across coding, search, tool use, and workflow execution, enabling scalable learning from complex agent-environment interactions. Together, Ling-2.6 and Ring-2.6 provide a practical pathway toward efficient, scalable, and open agentic systems. We open-source all checkpoints in the 2.6 family to support further research and development in practical agentic intelligence.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15007unread
Nemotron 3 Ultra: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning
NVIDIA (Allan), : (Allan), Aaron Blakeman (Allan), Aaron Thomas (Allan), Aastha Jhunjhunwala (Allan), Abhibha Gupta (Allan), Abhinav Khattar (Allan), Adam Rajfer (Allan), Adi Renduchintala (Allan), Adil Asif (Allan), Aditya Vavre (Allan), Adriana Flores Miranda (Allan), Ahmad Bilal (Allan), Aileen Zaman (Allan), Ajay Hotchandani (Allan), Akanksha Shukla (Allan), Akhiad Bercovich (Allan), Aleksander Ficek (Allan), Alex Gronskiy (Allan), Alex Kondratenko (Allan), Alex Steiner (Allan), Alex Ye (Allan), Alexander Bukharin (Allan), Alexandre Milesi (Allan), Ali Taghibakhshi (Allan), Alice Gatti (Allan), Alisa Liu (Allan), Alok Kumar (Allan), Amar Phanishayee (Allan), Ameya Sunil Mahabaleshwarkar (Allan), Amir Klein (Allan), Amit Zuker (Allan), Amnon Geifman (Allan), Anahita Bhiwandiwalla (Allan), Ananth Subramaniam (Allan), Andrea Santilli (Allan), Andrew Fulks (Allan), Andrew McHarg (Allan), Andrew Tao (Allan), Andrii Skliar (Allan), Anjulie Agrusa (Allan), Ankur Srivastava (Allan), Ankur Verma (Allan), Anna Shors (Allan), Anna Warno (Allan), Antoni-Joan Solergibert I Llaquet (Allan), Arham Mehta (Allan), Arkadiusz Nowaczynski (Allan), Arti Jain (Allan), Ashwath Aithal (Allan), Ashwin Poojary (Allan), Asif Ahamed (Allan), Asit Mishra (Allan), Asma Kuriparambil Thekkumpate (Allan), Atefeh Sohrabizadeh (Allan), Avinash Kaur (Allan), Avinash Vem (Allan), Ayush Dattagupta (Allan), Barath Subramaniam Anandan (Allan), Bardiya Sadeghi (Allan), Ben Lanir (Allan), Benedikt Schifferer (Allan), Besmira Nushi (Allan), Bilal Kartal (Allan), Bill Thiede (Allan), Bita Darvish Rouhani (Allan), Bo Deng (Allan), Bob Schatz (Allan), Boris Ginsburg (Allan), Boxin Wang (Allan), Brad Nemire (Allan), Brandon Norick (Allan), Brian Dang (Allan), Brian Westphal (Allan), Brian Yu (Allan), Brucek Khailany (Allan), Bryan Catanzaro (Allan), Carlo del Mundo (Allan), Caryln Aarish (Allan), Chankyu Lee (Allan), Chantal Hwang (Allan), Charbel Sakr (Allan), Charles Wang (Allan), Charlie Truong (Allan), Chen Cui (Allan), Cheng Cheng (Allan), Cheng-Ping Hsieh (Allan), Chenghao Zhang (Allan), Chenhui Deng (Allan), Chintan Patel (Allan), Chris Alexiuk (Allan), Christian Cosgrove (Allan), Christian Munley (Allan), Christine Harvey (Allan), Christopher Parisien (Allan), Chunyang Shen (Allan), Coco Li (Allan), Collin Neale (Allan), Cynthia Gao (Allan), Cyril Meurillon (Allan), Dan Gil (Allan), Dan Su (Allan), Dan Zhao (Allan), Dane Corneil (Allan), Daniel Afrimi (Allan), Daniel Egert (Allan), Daniel Korzekwa (Allan), Daniel Lo (Allan), Daniel Machlab (Allan), Daniel Serebrenik (Allan), Daniil Sorokin (Allan), Daria Gitman (Allan), Daria Levy (Allan), Darko Stosic (Allan), David Mosallanezhad (Allan), David Yu (Allan), Davit Karamyan (Allan), Deena Donia (Allan), Deep Debroy (Allan), Deepak Narayanan (Allan), Devin O'Kelly (Allan), Dheeraj Peri (Allan), Dhruv Nathawani (Allan), Di (Allan), Wu, Dima Rekesh, Divyanshu Kakwani, Donald Plummer, Dong Anh, Dongfeng Yu, Dongfu Jiang, Donnie Kim, Dorrin Poorkay, Duncan Riach, Dusan Stosic, Dustin VanStee, Eavan Meng, Edgar Minasyan, Edward Lin, Eileen Margaret Peters Long, Elad Sarafin, Elad Segal, Elena Lantz, Ellie Evans, Elliott Ning, Eric Chung, Eric Harper, Eric Pham-Hung, Eric Tramel, Eric Yang, Erick Galinkin, Erik Pounds, Erika Goncalves Goncalves, Evan Briones, Evan Wu, Evelina Bakhturina, Evgeny Tsykunov, Ewa Dobrowolska, Faisal Ladhak, Farzan Memarian, Fay Wang, Fei Jia, Felipe Soares, Felipe Vieira Frujeri, Feng Chen, Fengguang Lin, Ferenc Galko, Frank Sun, Frankie Siino, Frida Hou, Gal Hubara Agam, Gal Kaplun, Gantavya Bhatt, Gargi Prasad, Garvit Kulshreshtha, George Armstrong, Gerald Shen, Giulio Borghesi, Gordana Neskovic, Gorkem Batmaz, Grace Lam, Greg Mason, Greg Pauloski, Grigor Nalbandyan, Grzegorz Chlebus, Grzegorz Karch, Guan-Ting Liu, Guoming Zhang, Guyue Huang, Haggai Maron, Haifeng Qian, Haim Elisha, Haoxing Ren, Haran Kumar Shiv Kumar, Haribhau Hud, Harris Nover, Harrison Saturley Hall, Hayate Iso, Helen Ngo, Herbert Hum, Herman Sahota, Hexin Wang, Himanshu Soni, Hovhannes Tamoyan, Hua Li, Huanhuan Chen, Hui Li, Hui Wang, Huy Nguyen, Ian Chiles, Ido Galil, Ido Shahaf, Igor Gitman, Igor Shovkun, Ilya Loshchilov, Ingo Guehring, Itamar Schen, Itay Levy, Itay Neeman, Ivan Moshkov, Izik Golan, Izzy Putterman, Jaemin Choi, Jakub Slowikowski, Jan Kautz, Jane Polak Scowcroft, Jared Casper, Jatin Mitra, Jeffrey Glick, Jenny Chen, Jesse Oliver, Jiacheng Xu, Jiafan Zhu, Jialin Song, Jian Zhang, Jiantao Jiao, Jiaqi Zeng, Jie Lou, Jim King, Jimmy Zhang, Jingquan Wang, Jinhang Choi, Jinju Chu, Joey Conway, Joey Guman, Johan Jatko, Johannes Rausch, John Kamalu, John Roberts, Johnny Greco, Johnny Mensel, Jonah Alben, Jonas Yang, Jonathan Cohen, Jonathan Raiman, Joseph Jennings, Joshua Mabry, Joshua Pierce, Joyjit Daw, Julien Veron Vialard, Junkeun Yi, Jupinder Parmar, Kajal Jain, Kan Zhu, Kari Briski, Katherine Cheung, Katherine Luna, Keith Willowhawk, Keith Wyss, Keshav Santhanam, Kevin Shih, Kezhi Kong, Khanh Nguyen, Khushi Bhardwaj, Kirthi Shankar Sivamani, Konstantinos Krommydas, Krishna C. Puvvada, Krzysztof Pawelec, Kumar Anik, Kyle Keprios, Kylie Day, Lawrence McAfee, Leo Du, Leon Derczynski, Li Ding, Linda Liu, Lingjie Wu, Lior Kadoch, Lizzie Wei, Luis Vega, Luke Robison, Lun Su, Maarten Van Segbroeck, Maciej Jakub Mikulski, Maer Rodrigues de Melo, Magda Sypula, Mahan Fathi, Makesh Narsimhan Sreedhar, Makesh Tarun Chandran, Manoj Kilaru, Maor Ashkenazi, Marc Cuevas, Marc Romeijn, Marcin Chochowski, Mark Cai, Mark Mozolewski, Markus Kliegl, Marta Stepniewska-Dziubinska, Martyna Patelka, Mattei Machczynski, Matvei Novikov, Mauricio Ferrato, Maximilian Golub, Mehrzad Samadi, Melissa Corpuz, Mengru Wang, Mengxi Wu, Meredith Price, Meriem Boubdir, Micah Schaffer, Michael Andersch, Michael Boone, Michael Gschwind, Michael Lightstone, Michael Loh, Michal Bien, Michal Zawalski, Michelle Gill, Miguel Martinez, Mikail Khona, Mike Chrzanowski, Mike Houston, Mingyuan Ma, Minseok Lee, Mohamed Fawzy, Mohammad Dabbah, Mohammad Shoeybi, Mostofa Patwary, Nabin Mulepati, Najeeb Nabwani, Namit Dhameja, Narimane Hennouni, Natalie Hereth, Nathaniel Pinckney, Nave Algarici, Nave Assaf, Netanel Haber, Nicholas Knight, Nick Reamaroon, Nickson Quak, Nidhi Bhatia, Nikhil Desai, Nikolai Ludwig, Nima Tajbakhsh, Ning Xu, Nir Ailon, Nirmal Juluru, Nitin Nitin, Ofri Masad, Oleg Rybakov, Oleksii Hrinchuk, Oleksii Kuchaiev, Olivia Viessmann, Olivier Delalleau, Oluwatobi Olabiyi, Omer Ullman Argov, Omri Puny, Oren Tropp, Pablo Ribalta, Pallab Bhattacharya, Panos Lampropoulos, Parth Mannan, Pasha Shamis, Patrick Legresley, Paul Gibbons, Pavlo Molchanov, Pawel Morkisz, Peter Dykas, Peter Jin, Pierre-Yves Aquilanti, Pinky Xu, Piotr Januszewski, Piotr Laskiewicz, Pooya Jannaty, Prakash Gurumurthy, Pranav Prashant Thombre, Prasoon Varshney, Pritam Gundecha, Przemek Tredak, Puhui Meng, Qiyu Wan, Rabeeh Karimi Mahabadi, Rachel Oberman, Rachit Garg, Radha Sri-Tharan, Rahul Kandu, Rakshit Sanadhya, Ran El-Yaniv, Ran Zilberstein, Rasoul Shafipour, Ray Macalisang, Rayen Tian, Reka Kovacs, Renjie Pi, Rick Izzo, Rima Shahbazyan, Rishabh Garg, Rishi Puri, Rita Fernandes Neves, Ritchie Zhao, Ritika Borkar, Ritu Gala, Riyad Islam, Robert Clark, Robert Hesse, Robert Kirby, Roger Waleffe, Rohit Watve, Roi Koren, Ron Banner, Ruoxi Zhang, Russell J. Hewett, Ryan Prenger, Ryan Stewart, Ryota Egashira, Sadegh Mahdavi, Saee Paliwal, Sagar Singh, Sahil Modi, Salika Dave, Samantha Shinagawa, Samuel Kriman, Sandip Bhaskar, Sangkug Lym, Sanjay Kariyappa, Sanjeev Satheesh, Saran Vikas Murari, Satish Pasumarthi, Saurabh Mishra, Saurav Muralidharan, Scott Hara, Sean Narentharen, Selvaraj Anandaraj, Seonjin Na, Seonmeyong Bak, Seonmyeong Bak, Sepehr Sameni, Seph Mard, Serge Panev, Seth Henneman, Seth Poulos, Shahar Mor, Shantanu Acharya, Shaona Ghosh, Sharath Turuvekere Sreenivas, Sharon Mendelson, Shaun Kotek, Shawn Wang, Shay Aharon, Shaya Gharghabi, Sheng-Chieh Lin, Shi Chen, Shiqing Fan, Shirish Baskaran, Shreya Gopa, Shrimai Prabhumoye, Shubham Pachori, Shubham Toshniwal, Shuoyang Ding, Shwetha Krishnamurthy, Siddharth Singh, Simeng Sun, Sirshak Das, Sivakumar Arayandi Thottakara, Smita Ithape, Somshubra Majumdar, Soumye Singhal, Sri Harsha Singudasu, Sridhar Bhuvanapalli, Srimukh Veccham, Stas Sergienko, Stefania Alborghetti, Stephen Ge, Su Rong, Sugam Dipak Devare, Sukrit Rao, Sumeet Kumar Barua, Sungsoo Ha, Sunny Gai, Suriya Gunasekar, Suseella Panguluri, Suyog Gupta, Sviataslau Hinzburh, Sweta Priyadarshi, Syeda Nahida Akter, Talor Abramovich, Tan Bui, Tanay Varshney, Tatevik Ter-Hovhannisyan, Teodor-Dumitru Ene, Terry Kong, Thanh Do, Tianhe Zhang, Tiffany Moore, Tijmen Blankevoort, Tim Moon, Tiyasa Mitra, Tom Balough, Tomasz Grzegorzek, Tomasz Hliwiak, Tomer Asida, Tomer Bar Natan, Tomer Keren, Tomer Ronen, Tony Salim, Tony Wang, Traian Rebedea, Tugrul Konuk, Twinkle Vashishth, Udi Karpas, Ushnish De, Vahid Noorozi, Venkat Srinivasan, Venmugil Elango, Vibhor Agrawal, Victor Cui, Vijay Korthikanti, Vikas Mehta, Vinay Rao, Virginia Wu, Vitaly Kurin, Vitaly Lavrukhin, Vladimir Anisimov, Vu Pham, Wanli Jiang, Wasi Uddin Ahmad, Wataru Ishihara, Wei Du, Wei Ping, Weiheng Chai, Wenliang Dai, Wesley Helmholz, Will Jennings, Will Zhu, Wojciech Prazuch, Xiaowei Ren, Xiwen Yu, Yan Breek, Yang Chen, Yang Yu, Yangyi Chen, Yaniv Galron, Yashaswi Karnati, Yejin Choi, Yev Meyer, Yi-Fu Wu, Yian Zhang, Ying Lin, Yonatan Geifman, Yonggan Fu, Youngeun Kwon, Yu Yao, Yugi Guvvla, Yuki Huang, Yunsheng Liu, Zach Moshe, Zachary Newell, Zhilin Wang, Zhiyu Li, Zhongbo Zhu, Zhuolin Yang, Zihan Liu, Zijie Yan, Zsolt-Alon Wertheimer · 2026-06-17
arXiv:2606. 15007v1 Announce Type: new Abstract: We introduce Nemotron 3 Ultra, a 550 billion total and 55 billion active parameter Mixture-of-Experts Hybrid Mamba-Attention language model.
Read next because Nemotron 3 Ultra: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, source, token, recipe, compare, control, trained, length. Source: arxiv cs.CL (NLP).
arXiv:2606.15007v1 Announce Type: new Abstract: We introduce Nemotron 3 Ultra, a 550 billion total and 55 billion active parameter Mixture-of-Experts Hybrid Mamba-Attention language model. We pre-trained Nemotron 3 Ultra on 20 trillion text tokens, then extended the context length to 1M tokens, and post-trained using Supervised Fine Tuning (SFT), Reinforcement Learning (RL), and Multi-teacher On-Policy Distillation (MOPD). Nemotron 3 Ultra is our most capable model yet, employing multiple key technologies - LatentMoE, Multi Token Prediction (MTP), NVFP4 pre-training, multi-environment RLVR, MOPD, and reasoning budget control. Nemotron 3 Ultra achieves up to ~6x higher inference throughput as compared to state-of-the-art publicly available LLMs while attaining on-par accuracy. The state-of-the-art accuracy, high inference throughput, and 1M token context length make Nemotron 3 Ultra ideal for long-running autonomous agentic tasks. We open-source the base, post-trained, and quantized checkpoints, along with the training data and recipe on HuggingFace.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.14961unread
CoRA: Confidence-Rationale Alignment for Reliable Chain-of-Thought Reasoning
Juming Xiong, Weixin Liu, Kevin Guo, Congning Ni, Junchao Zhu, Chongyu Qu, Chao Yan, Katherine Brown, Avinash Baidya, Xiang Gao, Bradley Malin, Zhijun Yin · 2026-06-17
arXiv:2606. 14961v1 Announce Type: new Abstract: Chain-of-thought (CoT) reasoning can improve LLM performance, but high answer confidence may be misleading when the accompanying CoT rationale is plausible yet incomplete or poorly supported.
Read next because CoRA: Confidence-Rationale Alignment for Reliable Chain-of-Thought Reasoning overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: rect, alignment, correct, rate, compare, without, chain, model. Source: arxiv cs.CL (NLP).
arXiv:2606.14961v1 Announce Type: new Abstract: Chain-of-thought (CoT) reasoning can improve LLM performance, but high answer confidence may be misleading when the accompanying CoT rationale is plausible yet incomplete or poorly supported. We study confidence--rationale alignment: whether a model's confidence in its committed answer is justified by its generated rationale. We introduce a GRPO-based reinforcement learning framework that jointly rewards answer correctness, committed-answer probability, and rubric-based rationale support, where the rubric assesses grounding, coherence, task match, and connection to the selected answer without revealing the gold answer to the judge. Across MedQA, MathQA, and OpenBookQA using three open-weight LLMs, our method reduces the confidence--rationale alignment error by up to 26.51% compared with untuned checkpoints, SFT, and correctness-only GRPO, while maintaining competitive accuracy and often improving calibration. These results show that reliable CoT reasoning requires not only confident answers, but rationales that substantively support them.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.14943unread
Simplifying the Modeling of Arbitrary Conditionals in Natural Language
Yinhan Lu, Eric Elmoznino, L\'eo Gagnon, Sarthak Mittal, Tejas Kasetty, Guillaume Lajoie · 2026-06-17
arXiv:2606. 14943v1 Announce Type: new Abstract: Causal Transformers model sequences through an autoregressive factorization of the joint distribution, which enables efficient left-to-right decoding and conditional likelihood computation.
Read next because Simplifying the Modeling of Arbitrary Conditionals in Natural Language overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, text, eval, token, line, without, factor, contexts. Source: arxiv cs.CL (NLP).
arXiv:2606.14943v1 Announce Type: new Abstract: Causal Transformers model sequences through an autoregressive factorization of the joint distribution, which enables efficient left-to-right decoding and conditional likelihood computation. However, they cannot tractably sample from or evaluate arbitrary conditionals -- e.g., a block of text conditioned on past and future tokens. Recent work aims to solve this problem through novel architectures, but they often lead to sub-optimal modeling of such conditionals and degraded generations. We propose Arbitrary Conditionals GPT (AC-GPT) which introduces a simple modification to standard causal Transformers to enable evaluating and sampling from arbitrary conditionals -- including past, future, and mixed contexts -- within a single forward pass. Unlike prior approaches, our method preserves the standard left-to-right ordering and next-token prediction objective essential for both strong performance and efficient training on natural language. Crucially, this compatibility allows existing LLMs to be fine-tuned for arbitrary conditioning. Our empirical results indicate that our method outperforms baselines on modeling arbitrary conditionals, without degrading standard left-to-right performance.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.14875unread
Context Compression Is Not One Thing: Readable Symbolic Re-expression vs. Coherent Summary at Matched Budget
Sisong Bei, Mikhail L. Arbuzov, Ziwei Dong, Dmitri Kalaev, Alexey Shvets · 2026-06-17
arXiv:2606. 14875v1 Announce Type: new Abstract: We study context compression for multi-hop question answering with small language models.
Read next because Context Compression Is Not One Thing: Readable Symbolic Re-expression vs. Coherent Summary at Matched Budget overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, text, token, line, control, does, language, model. Source: arxiv cs.CL (NLP).
arXiv:2606.14875v1 Announce Type: new Abstract: We study context compression for multi-hop question answering with small language models. We propose Telegraph English, a readable symbolic format that rewrites retrieved passages into structured entity-relation statements, preserving reasoning evidence at lower token cost. In controlled experiments on MuSiQue, TwoWiki, and HotpotQA, Telegraph English outperforms three matched-budget compression baselines (character-level deletion, truncation, and random sub-sampling) on every dataset, with gains of 13 to 20 F1 percentage point. It also outperforms a coherent prose summary produced by the same encoder on the hardest dataset. A pre-registered depth-interaction hypothesis is null: the advantage does not grow with reasoning depth within datasets. We interpret these results as evidence that readable symbolic re-expression preserves entity content more densely than either natural language or coherent summarization at matched token budget.
source ↗open
score 100arxiv cs.LG (Machine Learning)arxiv:2606.17445unread
Toward Controllable Catalyst Inverse Design via Large-Scale Autoregressive Pretraining
Dong Hyeon Mok, Jonggeol Na, Seoin Back · 2026-06-17
arXiv:2606. 17445v1 Announce Type: new Abstract: Inverse design of heterogeneous catalysts remains challenging because catalyst surfaces exhibit substantial structural complexity with coupled surface-adsorbate interactions across a vast chemical space that is difficult to explore efficiently through conventional screening alone.
Read next because Toward Controllable Catalyst Inverse Design via Large-Scale Autoregressive Pretraining overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: rect, line, rate, control, without, binding, alone, trained. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17445v1 Announce Type: new Abstract: Inverse design of heterogeneous catalysts remains challenging because catalyst surfaces exhibit substantial structural complexity with coupled surface-adsorbate interactions across a vast chemical space that is difficult to explore efficiently through conventional screening alone. Although machine learning-based high-throughput screening has accelerated catalyst discovery, its efficiency inevitably declines as the search space grows, motivating the development of generative models that can directly construct catalysts with target properties. Here, we present a conditional catalyst generative model based on the Generative Pretrained Transformer architecture with a numerical embedding layer that enables the generation of catalyst structures conditioned on both categorical and continuous properties within a single autoregressive framework. The model was pretrained on 133 million catalyst structures and subsequently fine-tuned on approximately 460,000 optimized structures with associated categorical properties and binding energies for conditional generation. The resulting model achieved 98% structural validity, 95% optimization validity, and high categorical condition fidelity, with a 93 % joint match rate for adsorbate type and composition. For binding energy conditioning, the match rate of approximately 20% represents a four-fold improvement over the baseline training distribution, and the generated distributions shift systematically toward the target values, enabling a 1.5 to 4-fold improvement in screening efficiency for reaction-targeted catalyst discovery without additional fine-tuning. These results show that large-scale autoregressive pre-training, combined with explicit property conditioning, provides a practical route toward controllable catalyst generation and accelerated catalysts discovery.
source ↗open
score 100arxiv cs.LG (Machine Learning)arxiv:2606.17399unread
The Discrete-Log Clock: How a Transformer Learns Modular Multiplication
Huu Danh Nguyen (Stanford University) · 2026-06-17
arXiv:2606. 17399v1 Announce Type: new Abstract: When small transformers grok modular multiplication, prior work reports that the learned embedding has a "dense" Fourier spectrum requiring all frequencies.
Read next because The Discrete-Log Clock: How a Transformer Learns Modular Multiplication overlaps with clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "Implement Chen et al. persona-vector extraction recipe and compare to project's centroid-difference recipe". Matching terms: wrong, rate, implement, trained. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17399v1 Announce Type: new Abstract: When small transformers grok modular multiplication, prior work reports that the learned embedding has a "dense" Fourier spectrum requiring all frequencies. This contrasts with modular addition, where only a sparse set of key frequencies suffices. We show this density is an artifact of analyzing in the wrong basis. The natural Fourier transform for multiplication is not the standard additive DFT but the multiplicative character transform, which decomposes functions on the multiplicative group $(\mathbb{Z}/p\mathbb{Z})^*$ into its irreducible representations. Applying this transform to a grokked transformer trained on $a \cdot b \bmod 113$, we find the embedding spectrum becomes highly sparse (Gini coefficient 0.58 vs. 0.07 in the additive basis) with only 4 key frequencies carrying significant energy. Furthermore, 96.9% of MLP neurons are cleanly tuned to a single multiplicative frequency, and neuron activation heatmaps reveal 2D-periodic structure when reordered by the discrete logarithm. These results demonstrate the transformer reduces multiplication to addition in discrete-log space, implementing a "Discrete-Log Clock" algorithm analogous to Nanda et al.'s Clock algorithm for addition. The methodology generalizes: matching the analysis basis to the algebraic structure of the task reveals interpretable structure where standard tools see noise.
source ↗open
score 100arxiv cs.LG (Machine Learning)arxiv:2606.17377unread
Performance-Driven Environment Abstraction with Multi-Timescale Learning
Yue Guan, Dipankar Maity, Panagiotis Tsiotras · 2026-06-17
arXiv:2606. 17377v1 Announce Type: new Abstract: We study performance-driven environment abstraction for decision-making in large Markov decision processes.
Read next because Performance-Driven Environment Abstraction with Multi-Timescale Learning overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: rect, line, rate, compare, control, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17377v1 Announce Type: new Abstract: We study performance-driven environment abstraction for decision-making in large Markov decision processes. Rather than preserving geometric or topological structure, we seek abstractions that directly optimize decision quality. We model abstraction as a controlled approximation obtained by aggregating the state space and enforcing a shared action distribution within each aggregated state. For a fixed partition, we establish a performance guarantee that separates value-function approximation error from the loss introduced by action sharing. Guided by this analysis, we develop a multi-timescale reinforcement learning framework that jointly adapts the policy and a tree-structured environment abstraction. The resulting algorithm refines and coarsens regions of the state space based on Q-value discrepancies, balancing performance against abstraction size and complexity. Empirical results demonstrate substantial state compression, improved sample efficiency, and faster replanning compared to actor-critic baselines.
source ↗open
score 100arxiv cs.LG (Machine Learning)arxiv:2606.17345unread
Counterfactual Optimization of Baseball Pitch Sequences and Estimation of Its Impact on Season-Level Statistics
Ryota Takamido, Hiroki Nakamoto · 2026-06-17
arXiv:2606. 17345v1 Announce Type: new Abstract: Although pitch sequencing is a central topic in baseball analytics, previous studies have primarily focused on optimizing the final pitch within a single plate appearance, leaving the role of preceding setup pitches and their impact on long-term season-level performance insufficiently examined.
Read next because Counterfactual Optimization of Baseball Pitch Sequences and Estimation of Its Impact on Season-Level Statistics overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, middle, rate, trained, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17345v1 Announce Type: new Abstract: Although pitch sequencing is a central topic in baseball analytics, previous studies have primarily focused on optimizing the final pitch within a single plate appearance, leaving the role of preceding setup pitches and their impact on long-term season-level performance insufficiently examined. To address these issues, this study conducted counterfactual analyses using MLB Statcast data. A Transformer-based machine-learning model was trained to predict whether a target pitch would result in an in-play outcome or swing-out. Counterfactual pitch sequences were then generated by replacing either the final pitch or the preceding setup pitch with alternative pitch types and locations while keeping the surrounding contextual information fixed. Optimal counterfactual selections were defined as those that minimized the predicted in-play probability, and their expected effects on pitchers' seasonal statistics were estimated using regression models linking model outputs to season statistics. The results suggest that the optimization of both final and setup pitches may substantially influence season-level performance, including improvements of more than 1.0 in K/9. The analyses also provided several practical insights, including velocity-band-specific effective locations, the importance of pitch commands, and the expansion of pitch-selection options through middle-velocity pitches. These findings quantitatively support the strategic importance of pitch sequencing in baseball.
source ↗open
score 100arxiv cs.LG (Machine Learning)arxiv:2606.17192unread
Constrained Diffusion Models with Primal-Dual Inference
Samar Hadou, Yigit Berkay Uslu, Alejandro Ribeiro · 2026-06-17
arXiv:2606. 17192v1 Announce Type: new Abstract: This paper develops constrained diffusion models with primal-dual inference (PDI) to sample from optimal distributions of entropy-regularized optimization problems with \emph{average} constraints.
Read next because Constrained Diffusion Models with Primal-Dual Inference overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)". Matching terms: eval, source, rate, trained, factor, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17192v1 Announce Type: new Abstract: This paper develops constrained diffusion models with primal-dual inference (PDI) to sample from optimal distributions of entropy-regularized optimization problems with \emph{average} constraints. We formalize constrained sampling in the Lagrangian dual domain, where the optimal distribution takes the form of a Gibbs distribution indexed by the optimal dual variable. Rather than estimating this dual multiplier before sampling and freezing it throughout generation, PDI jointly infers the optimal primal distribution and its parametrizing dual variable. Each reverse diffusion step denoises using the score field associated with the current multiplier and then updates the multiplier through dual ascent using the estimated constraint violation of the denoised samples. To enable this conditional score field, we train a single dual-conditioned score network over the family of Gibbs distributions induced by the dual variables encountered during inference. We prove that the time average of the dual variables generated along the inference trajectory converges to a neighborhood of the dual optimum and bound the effect of residual dual mismatch on the terminal distribution through schedule-dependent stability factors. We evaluate PDI on constrained sampling from a mixture of Gaussians, wireless resource allocation, and portfolio management.
source ↗open
score 100arxiv cs.LG (Machine Learning)arxiv:2606.17182unread
Verified Detection and Prevention of Concurrency Anomalies in Multi-Agent Large Language Model Systems
Sajjad Khan · 2026-06-17
arXiv:2606. 17182v1 Announce Type: new Abstract: Multi-agent LLM systems share state through memory stores, vector indices, and tool registries.
Read next because Verified Detection and Prevention of Concurrency Anomalies in Multi-Agent Large Language Model Systems overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: class, under, rate, chain, language, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17182v1 Announce Type: new Abstract: Multi-agent LLM systems share state through memory stores, vector indices, and tool registries. We model such sharing as long-running read-generate-write operations under deterministic-generation semantics -- the regime durable-execution engines enforce by deterministic replay -- and formalize four concurrency anomalies in TLA+: stale-generation, phantom-tool, causal-cascade, and tool-effect reordering, structural analogues of classical isolation anomalies, each with a TLC counter-example. The exclusion lattice over these anomalies is trivial; the contribution is the mechanically verified realizability and strict separation of one maximal chain within it, $L_0 \subsetneq \cdots \subsetneq L_4$, to our knowledge the first machine-checked consistency hierarchy for such runtimes. A development of 274 Verus obligations (zero assume, zero admit; trust base: two structural axioms and a mutex correspondence) proves the detectors sound and complete against the specifications and each runtime its avoidance set. Three deployed Rust runtimes realize L0-L1 (pessimistic locking, serializable snapshot isolation, default-SI), each verified against stale-generation and refined to its state machine; L2-L4 are exec-mode-verified with dependency-free prevention twins (A3, A6, A2: 0/1000 versus 1000/1000), and L2 is run live across three model families (A3 prevented in all 120 retracted sessions). We reproduce a silent lost update in ByteDance's deer-flow, formalizing its fix as a verified $L_0 \to L_1$ refinement, and exhibit tool-effect reordering in LangGraph's ToolNode on unmodified output, removed by an L3 commit-order sequencer. The verified detector, refinements, and realizability artifacts are the contribution; the phenomena and lattice are classical.
source ↗open
score 100arxiv cs.LG (Machine Learning)arxiv:2606.17120unread
Noise-Driven Escape from Metastable Phases explains Grokking in Deep Neural Networks
Ibrahim Talha Ersoy, Karoline Wiesner · 2026-06-17
arXiv:2606. 17120v1 Announce Type: new Abstract: Deep neural networks (DNNs) exhibit first order phase transitions under variations of the L2 regularization strength, with each transition marking the onset of a new learnable feature.
Read next because Noise-Driven Escape from Metastable Phases explains Grokking in Deep Neural Networks overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "#351 follow-up: broader-vocab position-0 sweep at T=1.0 + position-1 suffix isolation". Matching terms: under, line, rate, test, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17120v1 Announce Type: new Abstract: Deep neural networks (DNNs) exhibit first order phase transitions under variations of the L2 regularization strength, with each transition marking the onset of a new learnable feature. Below a critical regularization strength, all features are in principle learnable, but coexisting metastable states, separated by energy barriers, can trap the network and impede convergence. A strength of DNNs is their ability to generalize. But many open questions remain, among them the origin of so called grokking: the abrupt, delayed onset of generalization after prolonged apparent overfitting. We show for linear DNNs that grokking is consistent with hysteresis in first-order L2 phase transitions: using L2 regularization to engineer deliberate trapping, we demonstrate that a model in a low-accuracy metastable state escapes only when SGD noise drives it across an energy barrier, with escape times following Arrhenius scaling. We reproduce grokking-like delayed convergence across two orders of magnitude in escape time by deliberately trapping models in metastable phases. Using sparse sub-sampling we also reproduce the canonical grokking curve where test error eventually approaches the final training error. Our work suggests that the number of metastable states equals the number of learnable features -- one per singular value of the data covariance -- the potential for hysteresis grows naturally with task complexity. We provide evidence that the same mechanism likely operates in general nonlinear DNNs. Our results provide routes toward more efficient learning schemes.
source ↗open
score 100arxiv stat.ML (Machine Learning)arxiv:2502.18049unread
Recursive Learning Without Collapse: A Weighting-Based Stabilization Framework
Hengzhi He, Shirong Xu, Guang Cheng · 2026-06-17
arXiv:2502. 18049v5 Announce Type: replace Abstract: Recent studies identified an intriguing phenomenon in recursive generative model training known as model collapse, where models trained on data generated by previous models exhibit severe performance degradation.
Read next because Recursive Learning Without Collapse: A Weighting-Based Stabilization Framework overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)". Matching terms: under, eval, line, rate, without, trained, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2502.18049v5 Announce Type: replace Abstract: Recent studies identified an intriguing phenomenon in recursive generative model training known as model collapse, where models trained on data generated by previous models exhibit severe performance degradation. Addressing this issue and developing more effective training strategies have become central challenges in generative model research. In this paper, we investigate this phenomenon within a novel framework, where generative models are iteratively trained on a combination of newly collected real data and synthetic data from the previous training step. To develop an optimal training strategy for integrating real and synthetic data, we evaluate the performance of a weighted training scheme in various scenarios, including Gaussian distribution estimation, generalized linear models, and nonparametric estimation. We theoretically characterize the impact of the mixing proportion and weighting scheme of synthetic data on the final model's performance. Our key finding is that, across different settings, the optimal weighting scheme under different proportions of synthetic data asymptotically follows a unified expression, revealing a fundamental trade-off between leveraging synthetic data and model performance. In some cases, the optimal weight assigned to real data corresponds to the reciprocal of the golden ratio. Finally, we validate our theoretical results on extensive simulated datasets and a real tabular dataset.
source ↗open
score 100arxiv stat.ML (Machine Learning)arxiv:2411.08821unread
Conditional Local Importance by Quantile Expectations
Kelvyn K. Bladen, Adele Cutler, D. Richard Cutler, Kevin R. Moon · 2026-06-17
arXiv:2411. 08821v4 Announce Type: replace Abstract: Global variable importance measures are commonly used to interpret the results of machine learning models.
Read next because Conditional Local Importance by Quantile Expectations overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: class, latin, rect, eval, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2411.08821v4 Announce Type: replace Abstract: Global variable importance measures are commonly used to interpret the results of machine learning models. Local variable importance techniques assess how variables contribute to individual observations. Current, popular methods, including LIME and SHAP, provide useful measures of feature contribution in the prediction space, while leaving opportunities for improved characterization of local structure in the model loss space. Additionally, they are not natively adapted for multi-class classification problems. We propose a new model-agnostic method for calculating local variable importance, CLIQUE, that highlights locally dependent relationships, provides improved stability over permutation-based methods, and can be directly applied to multi-class classification problems. Simulated and real-world examples show that CLIQUE emphasizes locally dependent information, captures interaction behavior beyond what can be evaluated by correlations, and assigns zero importance in regions where the response is invariant to changes in variables.
source ↗open
score 100arxiv stat.ML (Machine Learning)arxiv:2606.18218unread
Finite-Time Queue Peak Laws in Stochastic Networks: Logarithmic Scaling After Geometric Thresholds
Hao Liang, Cheng Tang, Yunzong Xu · 2026-06-17
arXiv:2606. 18218v1 Announce Type: cross Abstract: We study finite-horizon queue peaks in generalized switches, a standard stochastic-network model in which many queues share constrained service resources.
Read next because Finite-Time Queue Peak Laws in Stochastic Networks: Logarithmic Scaling After Geometric Thresholds overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: rect, source, rate, project, without, trained, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2606.18218v1 Announce Type: cross Abstract: We study finite-horizon queue peaks in generalized switches, a standard stochastic-network model in which many queues share constrained service resources. Arrivals may be dependent, time-varying, and adapted to the past; the standing load condition is uniform interior slack, meaning the conditional mean arrival vector stays in a fixed contraction of the capacity region. We show that this slack reshapes the finite-time peak law for drift-minimizing scheduling policies such as MaxWeight. The square-root envelope that is sharp without slack persists only up to a geometry-dependent threshold; beyond that threshold, the running maximum grows only logarithmically with the horizon, both with high probability and in expectation. The mechanism is self-normalization: in the current queue direction, the projected fluctuation scale is normalized by the stabilizing drift scale. This removes capacity geometry from the logarithmic coefficient, while geometry remains in the threshold. Matching lower bounds show that both the logarithmic term and a geometric threshold are unavoidable. When finite-time state-space collapse is available, the threshold can be sharpened using local bottleneck geometry. For generalized input-queued switches, we obtain finite-time peak bounds with tight logarithmic coefficients. Simulations illustrate the two-phase envelope, local geometric refinements, and variance-sensitive improvements predicted by the theory.
source ↗open
score 100arxiv stat.ML (Machine Learning)arxiv:2606.18199unread
Conformal Prediction Intervals with Tail-Specific Guarantees
Simone Cuonzo, Nina Deliu · 2026-06-17
arXiv:2606. 18199v1 Announce Type: cross Abstract: This paper extends classical conformal frameworks for constructing prediction intervals with global marginal coverage $1-\alpha$ to intervals that provide explicitly calibrated guarantees for the upper and lower tails separately.
Read next because Conformal Prediction Intervals with Tail-Specific Guarantees overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: class, rect, alpha, rate, control. Source: arxiv stat.ML (Machine Learning).
arXiv:2606.18199v1 Announce Type: cross Abstract: This paper extends classical conformal frameworks for constructing prediction intervals with global marginal coverage $1-\alpha$ to intervals that provide explicitly calibrated guarantees for the upper and lower tails separately. Focusing on split conformal prediction, we first construct lower and upper one-sided conformal intervals that achieve marginal validity, and then derive the induced two-sided interval by intersection. Theoretical results prove both tail-specific and global marginal coverage of the induced two-sided interval. Results are presented first for the exchangeable setting, where coverage has finite-sample guarantees, and then for non-exchangeable data, where guarantees are asymptotic. Simulation studies show that the proposed approach achieves improved directional calibration relative to classical two-sided intervals, especially relevant in skewed data. Finally, the benefit of the proposed framework is showcased in a financial application, where one aims for return maximization while seeking strict control on the left tail.
source ↗open
score 100arxiv stat.ML (Machine Learning)arxiv:2606.17939unread
Understanding Long-Term Dynamics of Individual Metro Usage: A Hidden Semi-Markov State Framework with Survival Analysis
Bingxun Wang, Valeria Maria Urbano, Shan He, Yang Chen, Wei Liu, Zhibin Jiang, Piercesare Secchi · 2026-06-17
arXiv:2606. 17939v1 Announce Type: cross Abstract: Understanding how individual metro usage evolves over multi-year horizons is essential for transit planning and passenger retention.
Read next because Understanding Long-Term Dynamics of Individual Metro Usage: A Hidden Semi-Markov State Framework with Survival Analysis overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: rect, under, rate, length, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2606.17939v1 Announce Type: cross Abstract: Understanding how individual metro usage evolves over multi-year horizons is essential for transit planning and passenger retention. However, existing approaches typically characterize mobility patterns as static clusters or short-term variability, leaving the lifecycle dynamics of transit participation underexplored. This study proposes a state-based lifecycle modeling framework that integrates Hidden Semi-Markov Models (HSMM) with discrete-time survival analysis to characterize the evolution of individual metro mobility. The HSMM infers latent mobility states with explicit duration distributions and a transition matrix governing regime changes, while the survival component models exit and re-entry events via state-dependent hazard functions conditioned on mobility-state trajectories and behavioral history. Applied to four years of smart card data from the Shanghai metro system (2021-2024), the framework enables the identification of interpretable mobility states, the characterization of transition dynamics, and the quantification of state-dependent exit and re-entry processes. The analysis reveals five robust mobility states with a directional transition hierarchy centered on an occasional-usage gateway state, and fundamentally different temporal mechanisms governing disengagement and return: exit hazard is state-dependent but duration-independent, whereas re-entry hazard decays sharply with inactivity length. These findings provide a methodological foundation for lifecycle-oriented mobility analysis and practical guidance for transit operators to identify at-risk users and time retention interventions.
source ↗open
score 100arxiv stat.ML (Machine Learning)arxiv:2606.17777unread
On Response-Adaptive Targeting Strategies for Multi-Treatment Experiments
Redouane Yagouti, R\'emy Degenne, Emilie Kaufmann · 2026-06-17
arXiv:2606. 17777v1 Announce Type: cross Abstract: Response-adaptive randomization (RAR) in clinical trials aims to improve ethical and statistical efficiency by dynamically allocating patients to treatments based on observed outcomes.
Read next because On Response-Adaptive Targeting Strategies for Multi-Treatment Experiments overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, text, alpha, rate, lora. Source: arxiv stat.ML (Machine Learning).
arXiv:2606.17777v1 Announce Type: cross Abstract: Response-adaptive randomization (RAR) in clinical trials aims to improve ethical and statistical efficiency by dynamically allocating patients to treatments based on observed outcomes. While RAR based on a target optimal allocation have been extensively studied for two-arms settings, their extension to multi-treatment experiments ($K \geq 2$) remains theoretically fragmented, with most existing methods focusing on specific algorithms or restricted target allocations. In this paper, we introduce a unified framework for response-adaptive targeting, the $\alpha$-Rebalancing Targeting Strategies ($\alpha$RTS), which generalize the ERADE two-armed strategy of Hu et al. [2009]. We prove that all designs in this family share fundamental asymptotic properties: strong consistency, asymptotic normality of allocation proportions and treatment effect estimators, and asymptotic efficiency. To address sparse target regimes (where some treatments are asymptotically eliminated), we further propose $\alpha$RTS with Forced Exploration, a variant that guarantees infinite sampling for all treatments while preserving the asymptotic guarantees. Extensive simulations illustrate the finite-sample behavior of $\alpha$RTS variants in a 3-armed context, highlighting in particular the critical role of forced exploration in sparse settings.
source ↗open
score 100arxiv stat.ML (Machine Learning)arxiv:2606.17515unread
Anytime-valid Optimal Policy Identification
Daniel Molitor · 2026-06-17
arXiv:2606. 17515v1 Announce Type: cross Abstract: We develop an anytime-valid framework for optimal policy identification from logged contextual bandit data.
Read next because Anytime-valid Optimal Policy Identification overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, class, line, rate, control, without, candidate. Source: arxiv stat.ML (Machine Learning).
arXiv:2606.17515v1 Announce Type: cross Abstract: We develop an anytime-valid framework for optimal policy identification from logged contextual bandit data. In many applied settings, the analyst wants to select the optimal policy from a candidate policy class $\Pi$, but data are generated by an externally determined logging policy that they do not control. The analyst may also wish to monitor evidence continuously, stopping once the optimal policy is clear rather than committing to a fixed sample size in advance. This paper addresses these challenges by constructing a time-indexed set $S_t$ that retains the true optimal policy set uniformly over time with high probability. The resulting procedure allows the analyst to monitor policy values, eliminate clearly suboptimal policies, and stop at data-dependent times without invalidating inference. When the optimal policy is unique, we define a stopping time for its identification and derive a sample-complexity bound scaling as $O\!\left(\frac{\log |\Pi|+\log\log(1/\Delta_{\min})}{\Delta_{\min}^2}\right)$, where $\Delta_{\min}$ is the gap between the best and second-best policy values. Simulations demonstrate that the anytime-valid approach can yield substantial sample savings relative to fixed-$N$ designs. An application to a large adaptive experiment on reducing misinformation online illustrates how the method provides a dynamic view as evidence on the optimal policy accumulates.
source ↗open
score 100arxiv stat.ML (Machine Learning)arxiv:2606.17383unread
Model Validation of Agentic AI Systems: A POMDP-Based Framework for Belief-State, Forecast, and Policy Validation
Matthew Francis Dixon · 2026-06-17
arXiv:2606. 17383v1 Announce Type: cross Abstract: Agentic artificial intelligence systems introduce a new class of model risk.
Read next because Model Validation of Agentic AI Systems: A POMDP-Based Framework for Belief-State, Forecast, and Policy Validation overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: class, under, rate, test, language, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2606.17383v1 Announce Type: cross Abstract: Agentic artificial intelligence systems introduce a new class of model risk. Unlike traditional predictive models, autonomous agents continuously acquire information, form beliefs regarding latent states of the environment, generate forecasts, select actions, and adapt their behavior over time. Existing validation methodologies focus primarily on predictive accuracy and therefore provide limited insight into the quality of the underlying decision process. This paper proposes a model validation framework for agentic AI based on Partially Observable Markov Decision Processes (POMDPs). The framework decomposes autonomous decision making into information, beliefs, forecasts, actions, and utility, allowing each component to be validated independently. Large language models (LLMs) are formalized as approximate Bayesian filtering operators, and a model-risk taxonomy is developed encompassing state-space, filtering, forecast, policy, utility-specification, and parameter risks. The model risk validation methodology is demonstrated through a portfolio-management case study in which an agent infers latent market regimes from market and macroeconomic information, generates belief-conditioned forecasts, and constructs portfolios using a Black--Litterman framework. Empirical validation combines performance analysis, belief calibration diagnostics, coverage tests, ablation studies, and parameter-sensitivity analysis. The results indicate that latent-state inference contributes independently to decision quality and that the principal conclusions remain robust across a broad range of parameter values. The principal contribution of the paper is a practical framework for extending established model risk management concepts to autonomous AI systems and providing a rigorous foundation for their validation, governance, and monitoring.
source ↗open
score 100arxiv stat.ML (Machine Learning)arxiv:2606.17364unread
A Polyak-Ruppert Central Limit Theorem for SA-Adam with Momentum and Non-Convergent Adaptive Preconditioning
Sunyoung An, Xiaoming Huo · 2026-06-17
arXiv:2606. 17364v1 Announce Type: cross Abstract: Adaptive optimizers combining preconditioning, momentum, and weight decay (Adam and AdamW) are, under Polyak-Ruppert averaging, candidate engines for one-pass inference.
Read next because A Polyak-Ruppert Central Limit Theorem for SA-Adam with Momentum and Non-Convergent Adaptive Preconditioning overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: class, under, alpha, line, rate, project, does, position. Source: arxiv stat.ML (Machine Learning).
arXiv:2606.17364v1 Announce Type: cross Abstract: Adaptive optimizers combining preconditioning, momentum, and weight decay (Adam and AdamW) are, under Polyak-Ruppert averaging, candidate engines for one-pass inference. Does the averaged iterate keep the classical Polyak-Ruppert central limit theorem (CLT), with sandwich covariance $H^{-1}SH^{-1}$ (Hessian $H$, gradient covariance $S$), under momentum and non-convergent preconditioning? The preconditioner-only analysis does not carry over: with momentum the canonical decomposition collapses to a tautology. Treating the augmented state (iterate, momentum buffer) as a time-varying linear stochastic approximation (SA), we prove (under local stabilization) positive drift stability, a non-autonomous Polyak-Ruppert CLT, and a projection identity. The upshot: the iterate-marginal covariance is exactly the plain stochastic gradient descent (SGD) sandwich $H^{-1}SH^{-1}$, so the adaptivity is asymptotically invisible. This holds for SA-Adam (sub-linearly vanishing momentum gain, $\gamma\in(\alpha,1)$; the sub-linear regime is essential), not constant-$\beta$ deployed Adam. Coupled $L_2$ weight decay yields the ridge-penalized sandwich, extending one-pass inference to regularized problems.
source ↗open
score 100arxiv stat.ML (Machine Learning)arxiv:2606.17308unread
Kernel-Based Functional Balancing for Causal Inference with Compositional Treatments
Sungbum Kim, Jiayi Wang · 2026-06-17
arXiv:2606. 17308v1 Announce Type: cross Abstract: We study causal effect estimation with compositional treatments, where the exposure lies on a simplex and the estimand is defined over compositions rather than scalar or binary values.
Read next because Kernel-Based Functional Balancing for Causal Inference with Compositional Treatments overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: rect, under, rate, project, without, position, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2606.17308v1 Announce Type: cross Abstract: We study causal effect estimation with compositional treatments, where the exposure lies on a simplex and the estimand is defined over compositions rather than scalar or binary values. By considering a projection of the average potential outcome onto the treatment space, a kernel-based covariate functional balancing approach is adopted for weight construction. The weights are obtained by directly minimizing a worst-case balancing error over a reproducing kernel Hilbert space (RKHS) defined on the joint space of treatments and covariates, instead of being estimated under a treatment assignment model. Building on these weights, an augmented weighted estimator (AWE) is proposed, where the outcome function is estimated via kernel ridge regression and combined with a marginal augmentation over the covariate distribution. Despite the complex structure of the resulting objective, a finite-dimensional convex optimization problem is formulated via a representer theorem and a low-rank approximation. The proposed estimator achieves $\sqrt{n}$-consistency without requiring consistent estimation or smoothness of the weights. An asymptotic normality result is established around a sample-specific target. Empirical performance is demonstrated through simulation studies and a real data application.
source ↗open
score 100arxiv stat.ML (Machine Learning)arxiv:2606.17215unread
Sum-of-Squares Degree Barriers for the Reweighted-Hinge Method in Robust Halfspace Learning: A Christoffel-Function Characterization
Xiaoyu Li · 2026-06-17
arXiv:2606. 17215v1 Announce Type: new Abstract: A certificate that removes outliers sees the data only through its low-degree moments, and an adversary exploits exactly this, hiding corruption where the clean data already looks typical, in the blind spot no bounded-degree test resolves.
Read next because Sum-of-Squares Degree Barriers for the Reweighted-Hinge Method in Robust Halfspace Learning: A Christoffel-Function Characterization overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)". Matching terms: class, under, source, rate, sees, test. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17215v1 Announce Type: new Abstract: A certificate that removes outliers sees the data only through its low-degree moments, and an adversary exploits exactly this, hiding corruption where the clean data already looks typical, in the blind spot no bounded-degree test resolves. That blind spot turns out to have an exact size: the Christoffel function of the clean marginal, the very quantity modern data analysis thresholds to detect outliers, here read from the adversary's side as the corruption a bounded-degree certificate cannot remove. We turn this inversion into the organizing principle of the reweighted-hinge approach to robustly learning $\gamma$-margin halfspaces under malicious noise (Shen, 2025; Zeng and Shen, 2025): the governing resource is the Sum-of-Squares degree of the outlier-removal certificate, and the resolution principle states that the maximal corruption mass which can hide at a center $c$ from a degree-$2t$ certificate is exactly the Christoffel function $\lambda_{t+1}(c)$ of the clean marginal. Three consequences follow, all against the certificate method (not information-theoretic). A margin-degree tradeoff: certifying the dense pancake to error $\epsilon$ costs SoS degree $\Omega(\log(1/\epsilon))$ or margin $\Omega(\sqrt{\log(1/\epsilon)}/\sqrt{d})$, explaining why the $\log(1/\epsilon)$ margin Shen (2025) records is forced, with a weighted-Chebyshev reduction making the threshold $2t=\Theta((|c|/s)^2)$ tight modulo one classical weighted-extremal estimate. A degree-$2$ outlier barrier: the resolution principle realized as an explicit instance on which degree $2$ is stuck at $\eta^{1/2}$ while degree $4$ escapes, locating the method's small breakdown rate in the degree, not the analysis. And a degree-$2t$ algorithm tracing the frontier $\eta^{1-1/2t}$ (recovering Shen (2025) at $t=1$), whose gain is an explicit constant, capped by the pancake density and shown unimprovable by the degree-$2$ barrier.
source ↗open
score 100arxiv stat.ML (Machine Learning)arxiv:2606.18074unread
Tensor-based second-order causal discovery
Nathan Ouyang, Kexin Wang, Anna Seigal · 2026-06-24
arXiv:2606. 18074v2 Announce Type: replace Abstract: Causal discovery seeks to uncover the causal dependencies among variables.
Read next because Tensor-based second-order causal discovery overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: rect, line, implement, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2606.18074v2 Announce Type: replace Abstract: Causal discovery seeks to uncover the causal dependencies among variables. For this purpose, we propose an algorithm called Tensor-based Second-order Causal Discovery (TSCD). Its input is a tensor obtained from the covariance matrices of observational and interventional data. Assuming the causal dependencies follow a linear structural equation model on a directed acyclic graph (DAG), TSCD outputs the DAG and the functions on its edges, requiring only that the noise variables are uncorrelated. We also implement a version of the approach for nonlinear models. Our focus on second-order statistics (via the covariance matrices) is motivated by their statistical and computational efficiency relative to higher-order moments, their identifiability relative to first-order statistics, and that they work regardless of whether the variables are Gaussian. We show that TSCD has identifiable causal order and parameters from a number of interventions that is logarithmic in the number of variables. Experiments show that TSCD is robust to noise, competitive with existing methods, and scales to hundreds of variables.
source ↗open
score 100arxiv stat.ML (Machine Learning)arxiv:2606.17995unread
Differential Privacy of Gaussian Process Posterior Sampling
Tomasz Maciazek · 2026-06-17
arXiv:2606. 17995v1 Announce Type: new Abstract: We study the privacy of releasing posterior sample paths from a Gaussian process (GP) when the entire training set including covariates and responses is private.
Read next because Differential Privacy of Gaussian Process Posterior Sampling overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, rate, leakage. Source: arxiv stat.ML (Machine Learning).
arXiv:2606.17995v1 Announce Type: new Abstract: We study the privacy of releasing posterior sample paths from a Gaussian process (GP) when the entire training set including covariates and responses is private. Unlike standard differential-privacy (DP) mechanisms that add external noise, posterior sampling is random by construction. We show that this intrinsic randomness yields DP guarantees by deriving explicit R\'enyi-DP bounds for GP posterior sample-path release. The bounds separate posterior-mean leakage from data-dependent posterior-covariance leakage showing that meaningful privacy depends sharply on effective ridge regularisation. We apply membership-inference attacks to show that empirical leakage follows the predicted dependence on regularisation, posterior variance and the number of released posterior sample-paths. Utility experiments on downstream posterior-sampling tasks identify noisy-observation regimes where privacy-compatible regularisation preserves useful decisions with modest utility loss. When stronger privacy is needed, the intrinsic guarantee can be sharpened by adding calibrated GP noise, providing an explicit additional privacy knob.
source ↗open
score 100arxiv stat.ML (Machine Learning)arxiv:2606.17319unread
Tight $L_\infty$ Sample Complexity for Low-Degree and Sparse Boolean Polynomials
Jasper van Doornmalen, Mathieu Molina, Victor Verdugo, Jos\'e Verschae · 2026-06-17
arXiv:2606. 17319v1 Announce Type: new Abstract: Motivated by the optimization of bounded binary black-box functions, we study the problem of learning polynomial surrogates over the Boolean hypercube.
Read next because Tight $L_\infty$ Sample Complexity for Low-Degree and Sparse Boolean Polynomials overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)". Matching terms: class, under, good, rate, control, factor. Source: arxiv stat.ML (Machine Learning).
arXiv:2606.17319v1 Announce Type: new Abstract: Motivated by the optimization of bounded binary black-box functions, we study the problem of learning polynomial surrogates over the Boolean hypercube. To ensure that optimizing the surrogate yields good solutions for the underlying objective, we require uniform $L_\infty$-error guarantees rather than the usual $L_2$-type guarantees. We characterize the minimax sample complexity of uniform estimation under subgaussian noise for two classes of bounded polynomials. First, for polynomials of degree at most $d$ on $n$ variables, the sample complexity scales as $n^{d+1}$. Second, for $s$-sparse Fourier-Walsh polynomials with $s \leq n$, it scales as $ns^2$. These rates differ structurally from the noiseless setting, where uniform exact recovery scales as $n^d$ and $ns$, respectively. Our lower bounds hold even for arbitrary adaptive learners, showing that the additional factors are intrinsic to the noisy cases. Standard Fourier-analysis tools for the $L_2$-norm do not naturally extend to the $L_\infty$-setting in a way that yields uniform guarantees. Our proofs overcome this difficulty by relying on suitably chosen auxiliary norms that serve as proxies for controlling the $L_\infty$-error. Together, our results provide a tight characterization of the sample complexity of learning optimization-safe polynomial surrogates.
source ↗open
score 98arxiv cs.LG (Machine Learning)arxiv:2606.17352unread
MM++: Unsupervised Scale-Invariant Multilayer OOD Detection via Top-K Gated Feature Fusion
Rahim Hossain, Md Tawheedul Islam Bhuian, Md Farhan Shadiq, Kyoung-Don Kang · 2026-06-17
arXiv:2606. 17352v1 Announce Type: new Abstract: We introduce MM++ (Multilayer Mahalanobis++), a fully unsupervised, strictly post-hoc, and scale-invariant framework for Out-of-Distribution (OOD) detection.
Read next because MM++: Unsupervised Scale-Invariant Multilayer OOD Detection via Top-K Gated Feature Fusion overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: class, full. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17352v1 Announce Type: new Abstract: We introduce MM++ (Multilayer Mahalanobis++), a fully unsupervised, strictly post-hoc, and scale-invariant framework for Out-of-Distribution (OOD) detection. To address the trade-off between scale invariance and hierarchical expressivity, MM++ constructs a principled joint feature space. It first identifies discriminative intermediate layers by measuring entropy density drops, which mark the boundaries of sharp semantic compression. By fusing these selected layers with the terminal representation, the framework captures latent cross-layer correlations while mitigating early-layer noise. Crucially, a Ledoit-Wolf regularized tied covariance matrix stabilizes this unified space, enabling reliable distance estimation. Requiring no auxiliary OOD data, classifier fine-tuning, or architectural modifications, MM++ delivers robust performance across distinct architectures for both near- and far-OOD detection.
source ↗open
score 94arxiv stat.ML (Machine Learning)arxiv:2606.17491unread
A Bayesian Boolean Matrix Factorization with Application to Copy Number Analysis in Cancer
Adolphus Wagala, Mehmet Samur, Giovanni Parmigiani · 2026-06-17
arXiv:2606. 17491v1 Announce Type: new Abstract: Binary data factorization is common, but real-valued methods ignore discreteness and yield hard-to-interpret factors.
Read next because A Bayesian Boolean Matrix Factorization with Application to Copy Number Analysis in Cancer overlaps with experiment "Follow-up to #354: cascading chunk-binding — does A→B, B→C, C→D propagate the full chain on a recipient trained only to emit A?", experiment "Factor screen for marker implantation + leakage (2^5: system-prompt length, answer-format length, persona-presence, on-policy, marker-only-loss)", experiment "#351 follow-up: broader-vocab position-0 sweep at T=1.0 + position-1 suffix isolation". Matching terms: full, factor, position, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2606.17491v1 Announce Type: new Abstract: Binary data factorization is common, but real-valued methods ignore discreteness and yield hard-to-interpret factors. Boolean Matrix Factorization (BooMF) instead decomposes a binary matrix into two lower-rank binary matrices via logical AND and OR, expressing the data as a Boolean disjunction of interpretable patterns. In cancer genomics, BooMF can reveal coordinated feature changes that may drive tumor evolution, unlike rotational or additive decompositions. Most existing BooMF methods are heuristic, greedy, sensitive to initialization, prone to local optima, and do not support principled model selection or uncertainty quantification. We introduce Bayesian Boolean Matrix Factorization (BBMF), a fully conjugate generative model with sparsity-inducing priors. It enforces Boolean constraints, yields interpretable latent factors with coherent uncertainty quantification, and admits Gibbs sampling with closed-form full conditionals. Because cancer evolution often involves widespread, near-simultaneous chromosome-number changes (e.g., whole-genome duplication followed by instability and selection), Boolean factorizations capture these patterns more naturally than additive models. Applied to arm-level copy-number alteration data in multiple myeloma, where entries indicate presence/absence of chromosomal-arm amplifications, BBMF finds a small set of interpretable bicliques linking patient subsets to recurrently co-altered chromosomal arms, providing a compact, biologically meaningful summary of tumor heterogeneity and demonstrating BBMF's utility for uncovering discrete latent structure in complex binary data.
source ↗open
score 94arxiv stat.ML (Machine Learning)arxiv:2606.17196unread
Another Look at Log-PCA for Probability Measures: A Dynamical Formulation and Statistical Convergence
Peng Xu, Changbo Zhu, Young-Heon Kim, Xiaohui Chen · 2026-06-17
arXiv:2606. 17196v1 Announce Type: new Abstract: This paper is concerned with learning principal variations of random probability measures on $\mathbb{R}^m$ under the Wasserstein geometry.
Read next because Another Look at Log-PCA for Probability Measures: A Dynamical Formulation and Statistical Convergence overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "#351 follow-up: broader-vocab position-0 sweep at T=1.0 + position-1 suffix isolation". Matching terms: under, line, rate, another. Source: arxiv stat.ML (Machine Learning).
arXiv:2606.17196v1 Announce Type: new Abstract: This paper is concerned with learning principal variations of random probability measures on $\mathbb{R}^m$ under the Wasserstein geometry. We introduce a new dynamical formulation to interpret the log-PCA, a linearized principal geodesic analysis, as a variational approach. Our differentiable version, termed as the Wasserstein Tangential PCA (WT-PCA), captures the local principal modes of geodesic variations of a (weighted) probability measure on the Wasserstein space via its covariance operator at barycenter. Based on the dynamical perspective and leveraging parallel transport structure of the optimal transport problems, we derive a general statistical convergence rate of the empirical WT-PCA when estimated from data in terms of the 2-Wasserstein distance between the population and empirical barycenter reference measures.
source ↗open
score 90arxiv cs.CR (Cryptography and Security)arxiv:2606.17533unread
SNAS: A Multi-Layer Defense-in-Depth Architecture for Secure Egress in Sandboxed Workloads
Niranjan Kumar Sharma, S Muralidhar, Samy Boshra-Riad, Mike Halcrow, Yuxiong He, Nitya Kumar Sharma, Shawn Xia, Haowei Yu, Elliott Brossard, Derek Denny-Brown, Choden Konigsmark, Bhanu Prakash, Brandon Baker, Andong Zhan · 2026-06-17
arXiv:2606. 17533v1 Announce Type: new Abstract: Snowpark enables data engineering and AI/ML workloads in Snowflake by executing user-defined functions in secure sandboxes.
Read next because SNAS: A Multi-Layer Defense-in-Depth Architecture for Secure Egress in Sandboxed Workloads overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: width, source, control. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.17533v1 Announce Type: new Abstract: Snowpark enables data engineering and AI/ML workloads in Snowflake by executing user-defined functions in secure sandboxes. Many of these workloads require external connectivity to access cloud APIs, external databases, or feature stores, creating a dependability challenge: how to provide transparent network access while preserving strict multi-tenant isolation and resource fairness. This paper presents Secure Network Access in Snowpark (SNAS), a production architecture for secure external communication from sandboxed workloads. SNAS combines Extended Berkeley Packet Filter (eBPF) packet filtering, Generic Network Virtualization Encapsulation (GENEVE) overlay networks, and distributed egress proxies for policy-driven egress control with low overhead. We describe the design, deployment, and measured production behavior of SNAS, including an eBPF-based bandwidth limiter using the Earliest Departure Time (EDT) algorithm, dual-tier policy enforcement, and safeguards for connection limiting and port exhaustion. SNAS is deployed across all Snowflake regions and supports large-scale production workloads including petabyte-scale data transfer and latency-sensitive external integrations.
source ↗open
score 90arxiv cs.CR (Cryptography and Security)arxiv:2606.17119unread
Graph neural networks at war: integrating cybersecurity and drone intelligence in the Israeli-Iranian conflict
Sozan Sulaiman Maghdid, Tarik Ahmed Rashid, Shavan Askar · 2026-06-17
arXiv:2606. 17119v1 Announce Type: new Abstract: Physical cyber systems have brought about new threats and challenges in detection and immediate response.
Read next because Graph neural networks at war: integrating cybersecurity and drone intelligence in the Israeli-Iranian conflict overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "#351 follow-up: broader-vocab position-0 sweep at T=1.0 + position-1 suffix isolation". Matching terms: under, rate, model. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.17119v1 Announce Type: new Abstract: Physical cyber systems have brought about new threats and challenges in detection and immediate response. This study examines how Graph Neural Networks (GNNs) can be used to aid cybersecurity and drone management in a physical cyber system comprising of cyber intrusions and unmanned aerial vehicles (UAVs). By providing a bridge between structural understanding of graphical neural networks, this work has provided an integrated procedure that allows intrusion detection systems to educate on underlying network structures, identify malicious activity, and facilitates drone response measures. Based on an emulation-based case study, cyberattacks models were created to provoke the responses of the drones, which proved that graph-based learning can assist with the situational awareness, swarm coordination, and adaptive maneuver. According to the performance valuation, this method has a detection rate of 94.2, average area under the receiver operating characteristic (ROC) of 0.955 and an average response time of 1.4 seconds. Comparative experiments reveal that proposed GraphSAGE network is more effective than the Graphical Convolutional Networks (GCNs) and Graphical Attention Networks (GATs) in the identical situation. Such findings prove that graphical neural networks can be used to avert intrusion and response of dynamic cyber-physical systems.
source ↗open
score 90arxiv cs.LG (Machine Learning)arxiv:2606.17451unread
Credibility-Weighted Pricing of Autonomous Vehicle Liability Under Operational Design Domain Shift
Doyeon Jang · 2026-06-17
arXiv:2606. 17451v1 Announce Type: new Abstract: Automated Driving System deployments create a foundational ratemaking challenge: sparse experience, shifting operational design domains, and non-stationary risk across software releases.
Read next because Credibility-Weighted Pricing of Autonomous Vehicle Liability Under Operational Design Domain Shift overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)". Matching terms: under, soft, rate. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17451v1 Announce Type: new Abstract: Automated Driving System deployments create a foundational ratemaking challenge: sparse experience, shifting operational design domains, and non-stationary risk across software releases. We propose a hierarchical Bayesian credibility framework pooling across cities, software versions, and territories via a learned ODD-similarity kernel, nesting Buhlmann-Straub as a limiting case. Demonstrated on 648 verified-engaged Waymo crashes across four U.S. metros from the NHTSA Standing General Order database against 116 million matched miles, city-aggregate credibility weights are moderate (0.12-0.46), partial pooling decisively outperforms no pooling, and a power analysis shows the learned kernel's advantage becomes detectable at approximately twelve deployed cities.
source ↗open
score 90arxiv stat.ML (Machine Learning)arxiv:2405.15379unread
Randomized Midpoint Method for Log-Concave Sampling under Constraints
Yifeng Yu, Shijie Zhang, Lu Yu · 2026-06-17
arXiv:2405. 15379v3 Announce Type: replace Abstract: In this paper, we study the problem of sampling from log-concave distributions supported on convex and compact sets, with a particular focus on the randomized midpoint discretization of both overdamped and kinetic Langevin diffusions in constrained domains.
Read next because Randomized Midpoint Method for Log-Concave Sampling under Constraints overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", experiment "Implement Chen et al. persona-vector extraction recipe and compare to project's centroid-difference recipe", experiment "Follow-up to #354: cascading chunk-binding — does A→B, B→C, C→D propagate the full chain on a recipient trained only to emit A?". Matching terms: under, project, trained. Source: arxiv stat.ML (Machine Learning).
arXiv:2405.15379v3 Announce Type: replace Abstract: In this paper, we study the problem of sampling from log-concave distributions supported on convex and compact sets, with a particular focus on the randomized midpoint discretization of both overdamped and kinetic Langevin diffusions in constrained domains. We revisit the proximal framework for handling constraints through projection operators and develop a more general formulation that encompasses Euclidean, Bregman, and Gauge projections. The resulting smooth approximation allows a unified and tractable analysis of Langevin algorithms and their variants under constraints. Within this framework, we establish convergence guarantees in Wasserstein-$q$ $(q\geqslant 1)$ distances between the smooth surrogate and the target distribution. We further derive complementary lower bounds, showing that the results are near-optimal in order. Building upon this tight approximation analysis, we obtain new convergence guarantees for the randomized midpoint Langevin algorithms and refined bounds for both vanilla and kinetic Langevin Monte Carlo methods under constraints, thereby advancing the theoretical understanding of constrained diffusion-based sampling.
source ↗open
score 90arxiv stat.ML (Machine Learning)arxiv:2606.17665unread
Non-asymptotic Tail Bounds for the Kostlan--Shub--Smale Field: Tensor PCA and Spherical $k$-Spin Complexity
Jean-Marc Aza\"is (IMT), Federico Dalmao (UDELAR), Yohann De Castro (ICJ, ECL, PSPM, IUF) · 2026-06-17
arXiv:2606. 17665v1 Announce Type: cross Abstract: This paper builds a hierarchy of explicit, non-asymptotic tail bounds for the supremum of the Kostlan--Shub--Smale (KSS) random field on the sphere, and applies it to two problems: Spiked Tensor PCA and the landscape of the spherical $k$-spin model.
Read next because Non-asymptotic Tail Bounds for the Kostlan--Shub--Smale Field: Tensor PCA and Spherical $k$-Spin Complexity overlaps with clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "Add C2 control arm (donor sees marker_B without marker_A) to disambiguate paired-marker binding from marker_B leaking alone", experiment "#351 follow-up: broader-vocab position-0 sweep at T=1.0 + position-1 suffix isolation". Matching terms: rate, control, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2606.17665v1 Announce Type: cross Abstract: This paper builds a hierarchy of explicit, non-asymptotic tail bounds for the supremum of the Kostlan--Shub--Smale (KSS) random field on the sphere, and applies it to two problems: Spiked Tensor PCA and the landscape of the spherical $k$-spin model. For Tensor PCA, we study the non-asymptotic statistical limits of estimating a rank-$R$ symmetric signal tensor of order~$k\ge 3$ and dimension~$d\ge 3$ from a single Gaussian observation at signal-to-noise ratio~$\lambda$, through the \emph{profile maximum likelihood estimator}, the MLE restricted to normalized rank-$R$ tensors of coherence at least~$\kappa$. Our analysis uses a single reduction: a deterministic geometric inequality (the Tube Method) and a rank-reduction step bound the estimation error by the supremum of the canonical KSS field, which the Kac--Rice formula turns into a Gaussian integral against the expected absolute characteristic polynomial of a shifted Gaussian Orthogonal Ensemble, controlled in turn by the four explicit tail bounds of our hierarchy (three from a Mehta--Fyodorov representation, one from a Ben Arous--Dembo--Guionnet large deviation). The same reduction yields two results, each with explicit constants. For estimation, a finite-$(k,d)$ error bound recovers the asymptotically optimal rate~$\sqrt{d\log k}$ of Perry, Wein and Bandeira, with explicit dependence on the rank~$R$ and the coherence~$\kappa$. For the landscape, a two-sided non-asymptotic bracketing of the annealed complexity of the spherical $k$-spin Hamiltonian recovers the Auffinger--Ben Arous--\v{C}ern\'y complexity function in the high-dimensional limit.
source ↗open
score 90arxiv stat.ML (Machine Learning)arxiv:2606.17267unread
Bayesian Poisson-Randomized Gamma Tensor Factorization with Application to International Trade Flows
Jie Jian, Aaron Schein · 2026-06-17
arXiv:2606. 17267v1 Announce Type: cross Abstract: We study sparse semi-continuous tensor data with excess zeros, heavy right tails, and slice-specific dispersion.
Read next because Bayesian Poisson-Randomized Gamma Tensor Factorization with Application to International Trade Flows overlaps with clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "Factor screen for marker implantation + leakage (2^5: system-prompt length, answer-format length, persona-presence, on-policy, marker-only-loss)", experiment "#351 follow-up: broader-vocab position-0 sweep at T=1.0 + position-1 suffix isolation". Matching terms: rate, factor, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2606.17267v1 Announce Type: cross Abstract: We study sparse semi-continuous tensor data with excess zeros, heavy right tails, and slice-specific dispersion. Such features arise naturally in monetary-valued multi-way data, such as international trade, where most exporter--importer--product--year cells are zero while positive values are continuous and highly variable. To model these data, we propose a Bayesian hierarchical tensor factorization model that places a low-rank CP structure on a latent Poisson rate tensor and couples it with a conditional Gamma model for positive outcomes, with rate parameters that can vary across slices within a mode. The model therefore separates the occurrence and magnitude of positive observations while borrowing strength across all tensor dimensions through a shared low-rank latent structure. To scale posterior inference to large arrays, we develop a hybrid variational--Monte Carlo algorithm that combines efficient coordinate ascent updates with a partially collapsed augmented-data sampler. Applied to approximately 60 million trade flows, the method surfaces multiway dependence across exporters, importers, products, and years that is difficult to recover from gravity-type or pairwise network analyses, which do not jointly model the product and temporal dimensions.
source ↗open
score 90arxiv stat.ML (Machine Learning)arxiv:2606.17185unread
Finsler Geometry, Graph Neural Networks, and You
T. Mitchell Roddenberry, Richard G. Baraniuk · 2026-06-17
arXiv:2606. 17185v1 Announce Type: new Abstract: Graph neural network architectures based on the graph Laplacian approximate the Laplace-Beltrami operator, thus limiting their application to isotropic operators.
Read next because Finsler Geometry, Graph Neural Networks, and You overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "Follow-up to #354: cascading chunk-binding — does A→B, B→C, C→D propagate the full chain on a recipient trained only to emit A?". Matching terms: under, line, trained. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17185v1 Announce Type: new Abstract: Graph neural network architectures based on the graph Laplacian approximate the Laplace-Beltrami operator, thus limiting their application to isotropic operators. As a nonlinear alternative to the Laplace-Beltrami operator, we consider estimates of the Finsler Laplacian on point clouds sampled from a manifold. We prove that these discrete estimates converge to the true operator on the manifold as the number of point samples grows. Moreover, we show that this operator can be expressed as a graph neural network layer, which we use to define a family of Finslerian graph neural networks constrained to express Finsler geometry. We show that Finslerian graph neural networks recover the geometry underlying nonlinear diffusion equations in practice.
source ↗open
score 74arxiv cs.LG (Machine Learning)arxiv:2606.17419unread
Generalization Guarantees for Multi-Input Neural Operator Learning in Sobolev Spaces
Yahong Yang, Zecheng Zhang, Wei Zhu, Wenjing Liao, Hao Liu · 2026-06-17
arXiv:2606. 17419v1 Announce Type: new Abstract: We develop approximation and generalization error estimates for multi-input neural operators, with the output error measured in Sobolev norms.
Read next because Generalization Guarantees for Multi-Input Neural Operator Learning in Sobolev Spaces overlaps with clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "#351 follow-up: broader-vocab position-0 sweep at T=1.0 + position-1 suffix isolation", experiment "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: rate, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17419v1 Announce Type: new Abstract: We develop approximation and generalization error estimates for multi-input neural operators, with the output error measured in Sobolev norms. In contrast to standard operator-learning settings with a single input function, our framework allows multiple input functions defined on possibly different domains, with different dimensions and Sobolev regularities. The derived rates explicitly quantify the contribution of each input space to the final error bound. In particular, in the balanced regime, the approximation and generalization rates are governed by the interaction between the input dimensions, regularities, and Sobolev orders, while the dependence on the model complexity retains a $\log\log/\log$-type structure. Our analysis provides a general theoretical framework for multi-input operator learning, including Sobolev training, and is applicable to operator learning problems arising from partial differential equations and scientific computing.
source ↗open
score 74arxiv stat.ML (Machine Learning)arxiv:2606.17260unread
Accelerated Convex Optimization via Hamiltonian Dynamics with Deterministic Integration Time
Xiuyuan Wang, Vishwak Srinivasan, Qiang Fu, Siddharth Mitra, Ashia Wilson, Andre Wibisono · 2026-06-17
arXiv:2606. 17260v1 Announce Type: cross Abstract: We develop Hamiltonian dynamics-based algorithms for smooth convex optimization that achieve accelerated rates of convergence.
Read next because Accelerated Convex Optimization via Hamiltonian Dynamics with Deterministic Integration Time overlaps with clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "Implement Chen et al. persona-vector extraction recipe and compare to project's centroid-difference recipe", experiment "#351 follow-up: broader-vocab position-0 sweep at T=1.0 + position-1 suffix isolation". Matching terms: rate, implement. Source: arxiv stat.ML (Machine Learning).
arXiv:2606.17260v1 Announce Type: cross Abstract: We develop Hamiltonian dynamics-based algorithms for smooth convex optimization that achieve accelerated rates of convergence. By exploiting contraction of averaged Hamiltonian flow trajectories rather than requiring contraction at trajectory endpoints, we show that Hamiltonian dynamics-based optimization methods admit deterministic and accelerated convergence guarantees, extending prior work that is limited to quadratic objectives or holds only in expectation. We analyze an idealized continuous-time algorithm and derive practical discrete-time implementations with optimal first-order complexity, thereby establishing Hamiltonian dynamics as a useful algorithmic primitive for deterministic accelerated convex optimization.
source ↗open
score 58arxiv cs.CR (Cryptography and Security)arxiv:2606.17116unread
Quantifying quantum risk: a measure of crypto agility
Coryan Wilson-Shah · 2026-06-17
arXiv:2606. 17116v1 Announce Type: new Abstract: Because of their ability to enable new forms of cryptanalysis, quantum computers pose a threat to the cryptographic algorithms that are widely used to secure contemporary computer systems.
Read next because Quantifying quantum risk: a measure of crypto agility overlaps with clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "#351 follow-up: broader-vocab position-0 sweep at T=1.0 + position-1 suffix isolation". Matching terms: rate. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.17116v1 Announce Type: new Abstract: Because of their ability to enable new forms of cryptanalysis, quantum computers pose a threat to the cryptographic algorithms that are widely used to secure contemporary computer systems. A practical quantum computer may emerge within the next ten years or so, but due to theorised "harvest now, decrypt later" style attacker behaviour, mitigations are necessary today. Recent advances in cryptography and security architecture show promise in supporting the design of systems that exhibit resilience against quantum-enabled cryptanalysis, however there is a key gap in the literature around the subject of deriving tolerances for such systems. In this paper, we introduce the concept of rotation time as a measure of crypto agility, and derive an approximation that links rotation time tolerance to security risk tolerance. Historical CVE data is used to calculate illustrative values for rotation time tolerance, which is found to be of the order of hours to days. This demonstrates that using crypto agility in conjunction with hybrid encryption is an effective approach for designing quantum-resilient systems, but may necessitate challenging technical and operational tolerances in order to meet organisational risk tolerances.
source ↗open

Threats and caveats

score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.18193unread
A Red-Team Study of Anthropic Fable 5 & Opus 4.8 Models
Nicola Franco · 2026-06-17
arXiv:2606. 18193v1 Announce Type: new Abstract: We evaluate the adversarial robustness of two frontier large language models (LLMs) developed by Anthropic, Fable 5 and Opus 4.
Read next because A Red-Team Study of Anthropic Fable 5 & Opus 4.8 Models overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, completions, under, eval, anth, rate, full, test. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.18193v1 Announce Type: new Abstract: We evaluate the adversarial robustness of two frontier large language models (LLMs) developed by Anthropic, Fable 5 and Opus 4.8, against four families of automated jailbreak attack across 7 826 harmful intents spanning a ten-category harm taxonomy. Using the HackAgent red-teaming framework, hundreds of thousands of adversarial attempts were generated and every apparent success was independently re-adjudicated by a panel of three judge models (majority vote). Both models resist the majority of attacks, but the residual surface is larger than aggregate framing suggests: it is dominated by adaptive iterative attacks, while static obfuscation is near-fully neutralised. The strongest adaptive search (tree-of-attacks) breaks Opus 4.8 on 11.5% of intents overall, whereas Fable 5 stays in the single digits (6.1% worst-case). Aggregate rates therefore should not be read as reassurance. Even in these hardened configurations, the two models produced 1 620 (Opus 4.8) and 702 (Fable 5) panel-confirmed harmful completions spanning every harm category, located automatically, cheaply, and within the first one or two refinement steps by an attacker model with no human expert in the loop. The reasonable conclusion is that even the best, most-tested frontier models remain reliably breakable under sustained automated pressure.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses robustness, adversarial.
source ↗open
score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.18190unread
Multi-Source Cybersecurity Logs: An ATT&CK-Labeled Dataset and SLM Evaluation
Abir Ashab Niloy, Ahmed Ryan, Imamul Hossain Rafi, Md Erfan, Md Rayhanur Rahman · 2026-06-17
arXiv:2606. 18190v1 Announce Type: new Abstract: Multi-stage cyberattacks span system, network, and browser logs.
Read next because Multi-Source Cybersecurity Logs: An ATT&CK-Labeled Dataset and SLM Evaluation overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: class, latin, under, eval, source, rate, compare, control. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.18190v1 Announce Type: new Abstract: Multi-stage cyberattacks span system, network, and browser logs. Detecting them requires correlating events across all three sources. Machine learning methods can learn these cross-source patterns, but they need labeled multi-source data. Existing public datasets fall short. Network-only datasets such as CICIDS and UNSW-NB15 miss host and browser activity. Host-focused datasets such as LMDG and CICAPT-IIoT lack browser telemetry. ATLAS includes all three sources but labels events only as malicious or benign, without MITRE Adversarial Tactics, Techniques, and Common Knowledge (ATT&CK) technique granularity. No public dataset combines all three sources with per-entry ATT&CK technique labels. We close the gap by building a multi-source log dataset of 870 sessions (70 attack, 800 benign) and approximately 2.3 million events. We captured system, network, and browser activity simultaneously on Windows endpoints. We labeled malicious events with ATT&CK technique IDs, covering 12 tactics and 53 techniques. We generated all attack data using real tools, including Remote Access Trojan (RAT), Command and Control (C2) tunnels, and cloud exfiltration. To demonstrate learnability, we fine-tuned three Small Language Models (SLMs) (Qwen2.5-1.5B, Llama-3.2-3B, Phi-4-Mini) using Low-Rank Adaptation (LoRA). We compared each against its base variant across ten metrics on two tasks: chunk classification and ATT&CK technique identification. Fine-tuning improved every model on every metric. Chunk classification accuracy rose from approximately 8% in the base variants to between 90% and 97% after fine-tuning. Technique identification remained challenging, with the best exact-match accuracy at 42%, although high partial-match scores show the models captured most of the underlying reasoning.
Potential threat/caveat for clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)": this item discusses adversarial, evaluation.
source ↗open
score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.18166unread
Evaluating Open-Source LLMs for Multi-Label ATT&CK Technique Classification on CTI Reports
Ahmed Ryan, Saad Sakib Noor, Md Erfan, Shaswata Mitra, Sudip Mittal, Md Rayhanur Rahman · 2026-06-17
arXiv:2606. 18166v1 Announce Type: new Abstract: Classifying Cyber Threat Intelligence (CTI) using MITRE Adversarial Tactics, Techniques, and Common Knowledge (ATT&CK) is essential for proactive defense, but historically required extensive human effort.
Read next because Evaluating Open-Source LLMs for Multi-Label ATT&CK Technique Classification on CTI Reports overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, class, under, eval, source, line, rate, language. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.18166v1 Announce Type: new Abstract: Classifying Cyber Threat Intelligence (CTI) using MITRE Adversarial Tactics, Techniques, and Common Knowledge (ATT&CK) is essential for proactive defense, but historically required extensive human effort. Pre-Large Language Model (LLM) automation sped up this process, but could not resolve the complex language and multi-step attack patterns found in unstructured CTI reports. LLMs addressed previous limitations by using contextual reasoning to understand unstructured text. However, current evaluations rely on simplified, single-technique sentences that ignore the complexity of real-world CTI reports, which often leads to inflated performance results. Consequently, the baseline performance of open-source LLMs on complex unstructured CTI reports remains unevaluated. To address this gap, we constructed a ground-truth dataset of 2,076 human-annotated sentences (1,281 technique-positive, 795 negative) from 83 complex unstructured CTI reports. These sentences were mapped to 114 unique ATT&CK techniques using a six-phase annotation process, achieving \k{appa} = 0.68 inter-annotator agreement. Using this dataset, we evaluated seven open-source LLMs ranging from 8B to 236B parameters across prompt strategy and temperature configurations. The highest-performing LLM achieved a micro-averaged F1 score of 0.22, establishing the empirical baseline for multi-label ATT&CK classification on complex unstructured CTI. Parameter size showed a statistically significant positive correlation with F1 score. Prompt strategy and temperature produced no statistically significant gains across model configurations. These results indicate that current open-source LLMs are insufficient for production-grade ATT&CK classification. The dataset, benchmark, and findings provide a reproducible foundation for future CTI research.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses limitation, limitations, negative, adversarial, evaluation.
source ↗open
score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.18052unread
An Empirical Analysis of AI Slop in Music Streaming
Stanley Wu, Josephine Passananti, Viresh Mittal, Wenxin Ding, Haitao Zheng, Ben Y. Zhao · 2026-06-17
arXiv:2606. 18052v1 Announce Type: new Abstract: Generative AI models lower the bar for content creation, making it easy for any user to create professional-looking images, text and music with minimal effort.
Read next because An Empirical Analysis of AI Slop in Music Streaming overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, line, rate, model. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.18052v1 Announce Type: new Abstract: Generative AI models lower the bar for content creation, making it easy for any user to create professional-looking images, text and music with minimal effort. This has enabled a new cottage industry around creation of "AI slop" mass quantities of mediocre content produced to generate revenue, often through misrepresentation as human-authored content, or scams involving automated scripts and fake consumption. While there are obvious parallels between the AI-slop industry and "traditional" email spam networks, it might be too early to determine if AI slop generation can grow into a similar self-sustaining industry. In this paper, we look specifically at the music industry, and explore the question: Can we prevent AI music slop from growing into a self-sustaining shadow industry? To answer this question, we characterize the current state of AI slop in music, and its pipeline from generation, distribution, and consumption by users on streaming platforms. By examining growth and engagement on Spotify, we confirm that AI music exhibits AI slop characteristics: the overwhelming majority (93%) of AI music receive few, if any listener plays, and are rarely recommended. AI musicians "spray and pray," releasing large volumes of music across multiple genres in hopes of generating a hit. We also explore the AI slop pipeline by generating and publishing our own AI tracks onto streaming through 11 indie music distributors. We find distributors have inconsistent and largely unenforced policies on AI music, making it surprisingly easy to publish mass produced AI songs. Finally, we consider AI music detection, and find that current methods lack accuracy or robustness. As generation costs decrease, we believe slop generation in music will become self-sustainable, unless concrete steps are taken by the music industry. We consider and discuss potential mitigation methods based on our findings.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses robustness.
source ↗open
score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.17986unread
ShellGames: Speculative LLM-Driven SSH Deception
Umberto Salviati, Fabio De Gaspari, Mauro Conti, Luigi Vincenzo Mancini · 2026-06-17
arXiv:2606. 17986v1 Announce Type: new Abstract: Cyber deception and Moving Target Defense are promising strategies that aim to disrupt adversaries by increasing uncertainty.
Read next because ShellGames: Speculative LLM-Driven SSH Deception overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: rect, under, correct, eval, line, rate, chain, trained. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.17986v1 Announce Type: new Abstract: Cyber deception and Moving Target Defense are promising strategies that aim to disrupt adversaries by increasing uncertainty. However, sustaining long-lived, credible interactive sessions with adversaries remains an open challenge. Large Language Models (LLMs) offer a promising path toward more dynamic deception systems, but suffer from key limitations that fundamentally limit their applicability, including: lack of persistent state, output inconsistencies, hallucinations, latency, and susceptibility to behavioral subversion that may reveal the deception. We propose ShellGames, an SSH shell simulator based on LLM designed to address these limitations. ShellGames combines five complementary techniques: (i) Automatic Chain-of-Thought and few-shot learning to improve correctness; (ii) memory management to maintain system state coherency; (iii) speculative command execution to reduce response latency; (iv) smart routing of complex interactive commands to a sandboxed environment; and (v) subversion detection leveraging the constrained input-output domain of shell environments. To enable systematic evaluation, we introduce a standardized benchmarking protocol and dataset spanning correctness, consistency, state tracking, and robustness tasks. ShellGames achieves $0.898$ command accuracy on correctness ($+5.3pp$ over baselines), $0.918$ sequence-level accuracy on consistency ($+36pp$), $0.98$ state tracking accuracy ($+18.3pp$), and $0.95$ accuracy on robustness ($+37pp$). A user study with $n=20$ participants confirms that ShellGames achieves realism comparable to a real shell under free exploration and outperforms traditional honeypots on perceived command coverage.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses limitation, limitations, robustness, evaluation, benchmark.
source ↗open
score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.17815unread
Beyond Native Success: Auditing Deployment-Interface Exposure of CLIP Backdoors
Kunlan Xiang, Haomiao Yang, Wenbo Jiang · 2026-06-17
arXiv:2606. 17815v1 Announce Type: new Abstract: Contrastive Language-Image Pre-training models are widely reused across downstream interfaces, including feature extraction, retrieval, reranking, and selection.
Read next because Beyond Native Success: Auditing Deployment-Interface Exposure of CLIP Backdoors overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, strong, text, fill, eval, extraction, control, does. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.17815v1 Announce Type: new Abstract: Contrastive Language-Image Pre-training models are widely reused across downstream interfaces, including feature extraction, retrieval, reranking, and selection. Existing CLIP backdoor, however, usually validate attacks on a small attack-native task, leaving unclear whether the same poisoned checkpoint remains exposed, weakens, or becomes not applicable when reused through other interfaces. We introduce DIFE, a Deployment-Interface Footprint Evaluation framework that audits backdoored CLIP checkpoints across deployment interfaces. DIFE makes various evaluations comparable by specifying each interface's component readout, trigger channel, target event, reference condition, and metric. DIFE also introduces effective-footprint diagnosis to identify the reusable CLIP component or component combination that carries exposure and explains where risk transfers. Auditing reproduced CLIP backdoors with DIFE reveals a structured landscape: native success is not a checkpoint-level risk certificate, exposure follows component footprints, text-side poisoning does not yield textual-encoder control, and some coupled attacks remain mechanism-bound. This audit reveals a import gapin existing CLIP backdoors: a textual encoder that itself becomes a reusable carrier of adversarial behavior. We therefore introduce BadTextTower to fill this gap. BadTextTower produces strong text-conditioned retrieval, reranking, and selection exposure while leaving visual-only reuse nearly clean.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses adversarial, evaluation.
source ↗open
score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.17562unread
Anywhere, Any-Stymie: Remote Activation of Trojan Malware on LiDAR with Modulated Signals
R. Spencer Hallyburton, Miroslav Pajic · 2026-06-17
arXiv:2606. 17562v1 Announce Type: new Abstract: LiDAR sensors are widely deployed in autonomous systems for 3D perception and safety-critical decision-making.
Read next because Anywhere, Any-Stymie: Remote Activation of Trojan Malware on LiDAR with Modulated Signals overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, eval, line, rate, implement, without, chain, test. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.17562v1 Announce Type: new Abstract: LiDAR sensors are widely deployed in autonomous systems for 3D perception and safety-critical decision-making. We identify a previously unexplored attack surface in which dormant malware embedded in the LiDAR sensing pipeline remains inactive during normal operation and can be externally triggered after deployment, without requiring access to sensor hardware or networking at attack time. To operationalize this threat, we design malware capable of low-level point-cloud manipulation and embed it into LiDAR firmware. This malware was developed in a closed research test environment with vendor technical support, rather than by exploiting an inherent production supply-chain vulnerability. To selectively trigger attack activation, we design and implement an optical trigger that remotely activates the malware by delivering a modulated signal into the sensing environment. Once triggered, the malware performs real-time point cloud manipulation, and we demonstrate false object injection and real object suppression on static and mobile victim platforms. Our evaluation first establishes attack feasibility, including static operation at 300~ft and recorded drive-by runs reaching 35~mph. We then illustrate quantitatively that injected person-like artifacts can remain semantically detectable by a state-of-the-art 3D object detector. Finally, we demonstrate multiple modes of safety-critical impact on a deployed tactical autonomous vehicle. Together, these results highlight the need for stronger integrity guarantees throughout the LiDAR sensor development and deployment pipeline.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation.
source ↗open
score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.17467unread
PARSE: Provenance-Aware Retrieval Sanitization for Professional Domain LLM Agents
Aaditya Pai · 2026-06-17
arXiv:2606. 17467v1 Announce Type: new Abstract: Prompt injection defenses evaluated on synthetic benchmarks do not generalize to real enterprise documents, which are longer, denser, and interleave legitimate authority language with factual content.
Read next because PARSE: Provenance-Aware Retrieval Sanitization for Professional Domain LLM Agents overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, class, rect, directive, eval, line, rate, language. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.17467v1 Announce Type: new Abstract: Prompt injection defenses evaluated on synthetic benchmarks do not generalize to real enterprise documents, which are longer, denser, and interleave legitimate authority language with factual content. We demonstrate this gap with a real-document benchmark of 122 tasks across five professional domains (financial, legal, medical, scientific, DevOps) using actual SEC filings, Federal Register rules, PubMed abstracts, arXiv papers, and GitHub postmortems. Paraphrasing, the strongest defense on synthetic benchmarks, shows no statistically significant attack success rate reduction on real documents (p=0.500) while degrading utility from 91.8% to 82.8%. We introduce PARSE (Provenance-Aware Retrieval Sanitization), a domain-aware, fact-preserving sanitization pipeline that classifies each sentence by injection likelihood, extracts structured facts before rewriting, and verifies fact preservation via a consistency-checking loop. A directiveness gate routes 59% of real enterprise documents to a lightweight path, concentrating computational cost on high-risk documents. PARSE achieves 15.6% attack success rate -- a 38% reduction versus the 25.4% baseline -- at 86.9% utility, the only condition that is both statistically significant (p=0.014, adequately powered) and maintains near-baseline utility. Practitioners should evaluate defenses on domain-matched real documents, not synthetic proxies.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses benchmark.
source ↗open
score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.17398unread
SoK: AI-Augmented Binary Reversing
Yujeong Kwon, Yiyue Zhang, Shakhzod Yuldoshkhujaev, Kexin Pei, Dokyung Song, Hyungjoon Koo · 2026-06-17
arXiv:2606. 17398v1 Announce Type: new Abstract: Binary reversing is fundamental to software understanding, vulnerability discovery, malware investigation, and firmware auditing.
Read next because SoK: AI-Augmented Binary Reversing overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)". Matching terms: under, soft, eval, line, rate, language, model. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.17398v1 Announce Type: new Abstract: Binary reversing is fundamental to software understanding, vulnerability discovery, malware investigation, and firmware auditing. However, it remains inherently challenging due to the irreversible loss of semantic information during compilation. Recent advances in machine learning, large language models (LLMs), and agentic AI systems have accelerated the adoption of AI-augmented binary reversing. Yet, the resulting body of work has become increasingly fragmented across reversing domains, artifact representations, learning approaches, and evaluation practices. This paper presents the first comprehensive systematization of knowledge on AI-augmented binary reversing. We analyze 144 research papers published since 2015, and organize them into 22 binary reversing domains according to the inference tasks. We further introduce a unified taxonomy spanning conventional and AI-augmented reversing pipelines. Our taxonomy connects traditional analysis techniques, binary-derived artifacts, representation strategies, learning paradigms, and downstream inference tasks, while clarifying the emerging roles of LLMs and agentic AI systems. By establishing a common vocabulary and structured framework, we provide a holistic view of the field's evolution over the past decade. Our study reveals common structures underlying seemingly disparate approaches, highlights persistent technical challenges and evaluation gaps, and identifies promising opportunities for future research. Collectively, these insights clarify the current state of the field and provide a foundation for the next generation of reliable and scalable AI-augmented binary reversing systems.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses evaluation.
source ↗open
score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.17283unread
ARVO: Atlas of Reproducible Vulnerabilities for Open-Source Software
Xiang Mei, Jordi Del Castillo, Pulkit Singh Singaria, Haoran Xi, Abdelouahab Benchikh, Tiffany Bao, Ruoyu Wang, Yan Shoshitaishvili, Adam Doup\'e, Hammond Pearce, Brendan Dolan-Gavitt · 2026-06-17
arXiv:2606. 17283v1 Announce Type: new Abstract: Achieving reproducibility, quantity, and diversity in vulnerability datasets has long been viewed as an inherent three-way trade-off, where improving one dimension often comes at the cost of the others.
Read next because ARVO: Atlas of Reproducible Vulnerabilities for Open-Source Software overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, rect, soft, eval, source, project, full. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.17283v1 Announce Type: new Abstract: Achieving reproducibility, quantity, and diversity in vulnerability datasets has long been viewed as an inherent three-way trade-off, where improving one dimension often comes at the cost of the others. In practice, reproducibility has been the dimension most often neglected. This has limited what can be automatically extracted from historical bug datasets, and has reduced their utility for downstream security research. In this work, we propose a method to produce a new security dataset which ensures reproducibility for diverse vulnerabilities at scale by identifying the key obstacles to large-scale bug reproduction and addressing them with general solutions. Using this method, we introduce full reproducibility to the largest open source software vulnerability dataset (OSS-Fuzz) and construct the ARVO dataset (an Atlas of Reproducible Vulnerabilities in Open-source software). ARVO is a large-scale dataset consisting of over 6,100 real-world vulnerabilities across 311 projects. Focusing on reproducibility, ARVO differs from existing datasets by providing each vulnerability in a form that can be consistently rebuilt, triggered, and analyzed across versions. Reproducibility also enables automatic identification of the corresponding patch for each vulnerability and supports direct interaction with vulnerabilities after code changes, capabilities that existing large-scale datasets do not provide. In our evaluation, ARVO successfully reproduces 81% of vulnerabilities and achieves 89.4% accuracy on the located patches. We also discuss ARVO's influence on both upstream practices and downstream security research.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation.
source ↗open
score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.17123unread
LineageMark: Multi-user White-box Watermarking for Contribution Tracing in Model Derivation Chains
Bingxue Zhang, Xiaofeng Xu, Feida Zhu · 2026-06-17
arXiv:2606. 17123v1 Announce Type: new Abstract: In open large language model (LLM) ecosystems, models are frequently adapted across multiple domains and applications, forming multi-stage derivation chains.
Read next because LineageMark: Multi-user White-box Watermarking for Contribution Tracing in Model Derivation Chains overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, under, eval, line, project, chain, stage, language. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.17123v1 Announce Type: new Abstract: In open large language model (LLM) ecosystems, models are frequently adapted across multiple domains and applications, forming multi-stage derivation chains. Consequently, tracking and verifying historical contributions is essential for model provenance and intellectual property protection. However, existing watermarking methods are mainly designed for single-user, one-time embeddings, often fail under repeated model derivation and incremental updates. To address this problem, we propose LineageMark, a multi-user white-box watermarking framework for model derivation chains. The framework encodes watermarks in model parameters using a projection-based approach. Stable carriers are first selected to reduce sensitivity to model changes, each watermark bit is then represented as a projection statistic over these carriers. Additional watermark insertions introduce only bounded perturbations in the projection space, and margin constraints are used to maintain signal integrity. We evaluate the effectiveness of LineageMark in multi-stage model derivation chains. Experimental results show that LineageMark preserves contributor watermarks across multi-stage derivation and supports incremental multi-user watermark insertion. Furthermore, it exhibits robustness against perturbations such as re-watermarking, fine-tuning, quantization, and pruning.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses robustness.
source ↗open
score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.17122unread
TrustErase: Auditable Instant Machine Unlearning with Passport-Embedded Representations
Rutger Hendrix, Leonardo G. Russo, Concetto Spampinato, Matteo Pennisi, Giovanni Bellitto · 2026-06-17
arXiv:2606. 17122v1 Announce Type: new Abstract: The demand for privacy-compliant AI has amplified the need for machine unlearning; yet, existing retraining or distillation-based methods remain unverifiable and computationally costly.
Read next because TrustErase: Auditable Instant Machine Unlearning with Passport-Embedded Representations overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: class, eval, without, position, model. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.17122v1 Announce Type: new Abstract: The demand for privacy-compliant AI has amplified the need for machine unlearning; yet, existing retraining or distillation-based methods remain unverifiable and computationally costly. We introduce TrustErase, a verifiable, data-free unlearning framework leveraging passport-embedded representations for instant, modular, and auditable forgetting. By treating passports as cryptographic keys within parameter-efficient adaptation layers, TrustErase enables the removal of specific classes or datasets through simple deactivation, without retraining, fine-tuning, or access to the original data. A singular value based decomposition conceals passports within model weights, ensuring that unlearning actions remain transparent and provably compliant. Evaluations on MNIST, CIFAR10 and CIFAR100 show that TrustErase matches or exceeds state-of-the-art benchmarks such as DELETE, L2UL, and Boundary Shrink, while operating in a strictly data-free regime. Ultimately, TrustErase establishes a new paradigm for trustworthy, accountable, and instantly forgettable AI systems.
Potential threat/caveat for clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)": this item discusses evaluation, benchmark.
source ↗open
score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.17114unread
An Evaluation of Data Leakage Risks in Tool-Using LLM Agents in Realistic Scenarios
Hankyul Baek, Jaewon Noh, Sang Seo, Yongsu Kim, Gabriel Waikin Loh Matienzo, Young Il Kim, Ee Wei Seah, Akriti Vij · 2026-06-17
arXiv:2606. 17114v1 Announce Type: new Abstract: AI agents are increasingly being adopted in enterprise and personal settings with access to emails, databases, documents, and other tools where they can read, update, and disseminate sensitive information.
Read next because An Evaluation of Data Leakage Risks in Tool-Using LLM Agents in Realistic Scenarios overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)". Matching terms: recipient, persona, rect, correct, eval, rate, full, leakage. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.17114v1 Announce Type: new Abstract: AI agents are increasingly being adopted in enterprise and personal settings with access to emails, databases, documents, and other tools where they can read, update, and disseminate sensitive information. Much of prior research on data leakage risks in agents has focused on adversarial data exfiltration through prompt injections and jailbreaks. However, sensitive information may also be exposed during non-adversarial use, creating leakage risks even when users issue benign requests. We report a joint evaluation by the Singapore AI Safety Institute and the Korea AI Safety Institute examining agent data leakage in 12 realistic, non-adversarial tasks spanning customer support, DevOps, web automation, and enterprise and personal productivity. The evaluation covers five risk types: lack of data awareness, audience awareness, policy compliance, data minimization, and access-boundary awareness. Both institutes tested a common set of scenarios mirroring real-world deployments using independent testing environments and task-specific LLM-judge rubrics. Across the three tested agents, none achieved fully correct and fully safe execution across all scenarios. Successful task completion often coincided with data-handling failures such as accessing unnecessary information or disclosing information to inappropriate recipients, indicating that capability and data-handling safety should be evaluated separately. Qualitative review also revealed claim-action mismatches, simulation-aware behavior, user-simulator role reversal, and interpretation gaps in automated judging. Overall, the results indicate that operational data leakage is a first-order agent-safety concern distinct from adversarial exfiltration and provide a methodology for future evaluations of agent data-handling safety.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses failure, failures, adversarial, evaluation.
source ↗open
score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.17111unread
Fractional Verkle Trees: A Hypertree Decomposition and Verified Proof Serialization Architecture for High-Performance Blockchain State Accumulators
Ekleen Kaur, Everton Fraga · 2026-06-17
arXiv:2606. 17111v1 Announce Type: new Abstract: Modern blockchain state management faces a critical scalability bottleneck: maintaining cryptographic commitments over hundreds of millions of entries becomes computationally prohibitive.
Read next because Fractional Verkle Trees: A Hypertree Decomposition and Verified Proof Serialization Architecture for High-Performance Blockchain State Accumulators overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)", experiment "Implement Chen et al. persona-vector extraction recipe and compare to project's centroid-difference recipe". Matching terms: width, implement, full, chain, position. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.17111v1 Announce Type: new Abstract: Modern blockchain state management faces a critical scalability bottleneck: maintaining cryptographic commitments over hundreds of millions of entries becomes computationally prohibitive. Ethereum's transition to Verkle Trees: polynomial commitment accumulators reducing proof sizes from O(width * depth) to O(depth) via constant-size IPA vector commitments, is a critical step toward stateless operation. Yet, current implementations exhibit pathological characteristics that burden home validators. We identify four inefficiencies in the reference go-verkle implementation \cite{kaur2025goverkle, kaur2025goethereum}: (1) phantom node creation during non-existent account deletion; (2) 64-byte database keys triggering excessive LSM-tree compaction; (3) redundant memory copying in proof deserialization; (4) a Proof of Absence wire format incompatibility causing non-deterministic serialization. We present Fractional Verkle Trees (FVT), a hypertree decomposition partitioning global state into N independent sub-accumulators coordinated by a Merkle commitment tree, achieving improved cache locality, zero-lock-contention goroutine-parallel commitment computation, and faster root recomputation (91 $\mu$s vs $\sim$500 ms). We address each inefficiency via existence checks, 32-byte SHA256 node references, zero-copy reference-counted buffers, and HashMap-based lexicographic deduplication. Benchmarks on Apple M1 Pro show 57\% heap allocation reduction (566,760 to 242,004 bytes per 10K proofs), parallel insertion at 2,433 ns/op, and network-wide elimination of 4.85 PB/year across 6,000 full nodes, advancing the Ethereum stateless roadmap.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses benchmark.
source ↗open
score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.17109unread
Timestamp-Aware Spatio-Temporal Graph Contrastive Learning for Network Intrusion Detection
Jianli Dai, Guangwei Wu, Jiacheng Li, Weiping Wang, An He, Xinjun Xiao · 2026-06-17
arXiv:2606. 17109v1 Announce Type: new Abstract: Given their effectiveness in modeling the relational structure among network traffic flows, graph neural networks (GNNs) have been widely adopted in network intrusion detection systems (NIDSs).
Read next because Timestamp-Aware Spatio-Temporal Graph Contrastive Learning for Network Intrusion Detection overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, rate, without, full, model. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.17109v1 Announce Type: new Abstract: Given their effectiveness in modeling the relational structure among network traffic flows, graph neural networks (GNNs) have been widely adopted in network intrusion detection systems (NIDSs). However, most existing GNN-based NIDS approaches focus on the relational structure of traffic flows, and treat them as temporally independent, which limits their ability to cope with evolving attack behaviors. Moreover, their reliance on supervised or semi-supervised learning often restricts generalization to unseen attacks. To address these limitations, we propose a novel self-supervised GNN-based framework. To the best of our knowledge, the proposed model is among the first self-supervised GNN-based NIDS models to explicitly leverage real timestamps, which provides faithful temporal dependencies for representation learning. We first construct a series of temporal graphs from network traffic flows according to their timestamps, and then employ an E-GraphSAGE and LSTM based encoder to fully extract temporal information and spatial dependencies of network traffic, without introducing time-costly attention mechanisms. A multi-view graph contrastive learning (GCL) scheme is introduced, where temporal, spatial, and feature contrasts are jointly performed to capture temporal continuity, preserve structural consistency, and improve the generalization and robustness of the learned representations, respectively. In addition, a gradient-norm-based adaptive weighting strategy is designed to optimize the contrastive loss weights. Experimental results on four representative NIDS datasets with real timestamps demonstrate that our method significantly outperforms existing self-supervised approaches and achieves performance comparable to the supervised state-of-the-art GNN method, while maintaining high computational efficiency.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses limitation, limitations, robustness.
source ↗open
score 100arxiv cs.CR (Cryptography and Security)arxiv:2606.17092unread
Securing Multi-Agent GIS Systems: Risk Evaluation and Prompt Hardening Optimization
Kyle Gao, Pranavi Kotta, Linlin Xu, Jonathan Li, David A. Clausi · 2026-06-17
arXiv:2606. 17092v1 Announce Type: new Abstract: Agentic systems are increasingly integrated with geographic information systems (GIS), where multi-agent coordination enables complex conversational and spatial analysis but introduces security risks.
Read next because Securing Multi-Agent GIS Systems: Risk Evaluation and Prompt Hardening Optimization overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "Add C2 control arm (donor sees marker_B without marker_A) to disambiguate paired-marker binding from marker_B leaking alone". Matching terms: eval, rate, without, test. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.17092v1 Announce Type: new Abstract: Agentic systems are increasingly integrated with geographic information systems (GIS), where multi-agent coordination enables complex conversational and spatial analysis but introduces security risks. This work presents a security-oriented framework for risk identification, evaluation, and mitigation in a multi-agent GIS system while maintaining adaptability to broader agentic architectures. We test the agentic system of a commercial geospatial partner while developing a modular state-machine-based orchestration framework that abstracts agent behavior into reusable components. We evaluate robustness using a red-teaming framework with an adaptive attacker LLM and a deterministic judge that produces binary outcomes with supporting rationales across multi-turn attacks. We further improve resilience with a prompt optimization framework that treats prompts as structured signatures and injects adversarial demonstrations, enabling systematic security improvements without degrading task performance.
Potential threat/caveat for clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)": this item discusses robustness, adversarial, evaluation.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17821unread
DecoSearch: Complexity-Aware Routing and Plan-Level Repair for Text-to-SQL
Esteban Schafir, Xu Zheng, Hojat Allah Salehi, Zhuomin Chen, Mo Sha, Wei Cheng, Dongsheng Luo · 2026-06-18
arXiv:2606. 17821v1 Announce Type: new Abstract: Large Language Models (LLMs) have demonstrated remarkable capabilities in translating natural language to SQL, yet existing methods still falter on complex queries requiring multi-step, data-aware reasoning.
Read next because DecoSearch: Complexity-Aware Routing and Plan-Level Repair for Text-to-SQL overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, latin, rect, token, line, rate, without, full. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17821v1 Announce Type: new Abstract: Large Language Models (LLMs) have demonstrated remarkable capabilities in translating natural language to SQL, yet existing methods still falter on complex queries requiring multi-step, data-aware reasoning. We introduce DecoSearch, a training-free framework that addresses this by routing each query to the appropriate level of reasoning effort. A lightweight Schema Selector first prunes the full database schema to the relevant tables and columns. An LLM Judger then decides whether the question requires decomposition: straightforward questions follow a direct generation path and complex ones are escalated to a Directed Acyclic Graph (DAG) of atomic sub-questions, each solved by a targeted SQL generation step. A RAG component grounds the decomposer with semantically similar training examples, and a Topology Refiner restructures the reasoning plan when execution failures signal a flawed decomposition rather than a fixable SQL error. DecoSearch achieves 70.53% execution accuracy on BIRD and 88.31% on Spider with a DeepSeek backbone, surpassing all training-free baselines while consuming an order of magnitude fewer tokens than competing methods. It also functions as a model-agnostic wrapper, consistently improving fine-tuned SQL generation backbones without any modification to the pipeline.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses failure, failures.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17735unread
Shattering the Autoregressive Curse: Dynamic Epistemic Entropy Orchestrated Erasable Reinforcement Learning for LLMs
Ziliang Wang, Kang An, Faqiang Qian, Jialu Cai, Cijun Ouyang, Yuhang Wang, Qibing Ren, Yichao Wu · 2026-06-18
arXiv:2606. 17735v1 Announce Type: new Abstract: Although reinforcement learning (RL) has expanded the cognitive boundaries of large language models (LLMs), it often remains vulnerable to the autoregressive curse in long-horizon logical reasoning: small epistemic perturbations introduced early in generation can propagate irreversibly along the Markov decision process flow, triggering cascading failures that drive the reasoning trajectory toward collapse.
Read next because Shattering the Autoregressive Curse: Dynamic Epistemic Entropy Orchestrated Erasable Reinforcement Learning for LLMs overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, line, rate, cascading, propagate, capability, lora, language. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17735v1 Announce Type: new Abstract: Although reinforcement learning (RL) has expanded the cognitive boundaries of large language models (LLMs), it often remains vulnerable to the autoregressive curse in long-horizon logical reasoning: small epistemic perturbations introduced early in generation can propagate irreversibly along the Markov decision process flow, triggering cascading failures that drive the reasoning trajectory toward collapse. To overcome this autoregressive cascade, in which a single early mistake can compromise all subsequent reasoning steps, we propose dynamic epistemic entropy orchestrated erasable reinforcement learning ($\text{E}^3\text{RL}$). $\text{E}^3\text{RL}$ eliminates reliance on external signals by grounding the model's endogenous local autoregressive cross-entropy as an intrinsic coordinate of epistemic uncertainty. By introducing segment-level adaptive dynamic thresholds and advantage allocation, $\text{E}^3\text{RL}$ enables the model to precisely excise localized logical defects while reusing historical key-value (KV) cache streams, thereby endowing the reasoning process with a self-healing capability. We train $\text{E}^3\text{RL}$ on the DeepMath-103k dataset. Experimental results show that $\text{E}^3\text{RL}$ reshapes the exploration efficiency of long-sequence reasoning and improves sample efficiency while maintaining linear memory overhead. On mathematical reasoning benchmarks such as AIME, $\text{E}^3\text{RL}$ achieves substantial performance gains, with the 4B and 8B parameter models surpassing previous state-of-the-art (SOTA) results by 5.349\% and 6.514\%, respectively. These findings suggest that $\text{E}^3\text{RL}$ shatters the autoregressive curse in long-sequence reasoning and establishes a theoretical and systems-level foundation for the next generation of self-healing artificial general intelligence (AGI).
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses failure, failures, benchmark.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17727unread
LongWebBench: Evaluating Structural and Functional Webpage Generation in Long-Horizon Settings
Yi Zhao, Zhen Yang, Mengpan Chen, Mingde Xu, Shanghui Gong, Xijun Liu, Jibing Gong, Jie Tang · 2026-06-18
arXiv:2606. 17727v1 Announce Type: new Abstract: Recent vision-language models (VLMs) have shown promising progress in generating webpages from visual inputs, yet existing evaluations mainly focus on short, single-screen, and largely static webpages.
Read next because LongWebBench: Evaluating Structural and Functional Webpage Generation in Long-Horizon Settings overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, under, eval, source, line, length, screen, language. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17727v1 Announce Type: new Abstract: Recent vision-language models (VLMs) have shown promising progress in generating webpages from visual inputs, yet existing evaluations mainly focus on short, single-screen, and largely static webpages. We introduce LongWebBench, a benchmark for evaluating long-horizon webpage generation from both structural and functional perspectives. LongWebBench contains 490 real-world long webpages for structural fidelity evaluation and 507 goal-oriented interaction tasks over 129 webpages for functional evaluation. It employs two complementary protocols: a multi-dimensional VLM-based metric for assessing long-range structural coherence, and a DOM-augmented agent-based pipeline for end-to-end functional verification. We further examine the automatic evaluation protocols through human agreement analysis. Experiments with state-of-the-art open-source and proprietary VLMs under single-image and multi-image settings reveal that structural fidelity degrades as webpage length increases, while visually plausible generations often fail to support executable multi-step interactions. These results highlight the need to evaluate long webpage generation beyond visual similarity, with executable interaction as a core criterion. Our code and data are available at https://github.com/zheny2751-dotcom/LongWebBench.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation, benchmark.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17698unread
EComAgentBench: Benchmarking Shopping Agents on Long-Horizon Tasks with Distributed Hidden Intent
Zeyao Du, Tong Li, Haibo Zhang · 2026-06-18
arXiv:2606. 17698v1 Announce Type: new Abstract: As LLM-based shopping agents enter production, existing benchmarks fail to capture how a shopper's requirements arrive: stated implicitly in the query, recorded in a profile, or revealed only when the right question is asked.
Read next because EComAgentBench: Benchmarking Shopping Agents on Long-Horizon Tasks with Distributed Hidden Intent overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, strong, text, eval, source, rate, full, candidates. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17698v1 Announce Type: new Abstract: As LLM-based shopping agents enter production, existing benchmarks fail to capture how a shopper's requirements arrive: stated implicitly in the query, recorded in a profile, or revealed only when the right question is asked. Benchmarks that expose full intent upfront and grade only the final choice can neither pose this long-horizon challenge nor explain which requirement an agent missed. To address this gap, we introduce EComAgentBench, a benchmark of 662 tasks grounded in real Amazon products and reviews. Each task scatters these requirements across a visible query, a tool-gated profile, and scripted clarification; an agent must uncover hidden intent, verify candidates against attributes and review evidence, and commit to a single product within 100 tool calls. Moreover, typed, source-tagged rubrics grade every task, attributing each failure to a requirement and its source. Construction is automated yet reliable, with every answer fixed in code before any text is generated and every sample validated. Our evaluation of seven models reveals that even the strongest attains only 57.1% overall accuracy, and rubric satisfaction degrades from visible to hidden sources. Overall, we believe EComAgentBench will serve as a reproducible foundation for moving shopping agents from single-query search toward dependable assistance over long horizons.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses failure, evaluation, benchmark.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17657unread
Using Cognitive Models to Improve Language Model Simulation of Human Persuasion Games
Zirui Cheng, Zeyu Shen, Thomas L. Griffiths, Peter Henderson · 2026-06-18
arXiv:2606. 17657v1 Announce Type: new Abstract: People make decisions differently in strategic interactions.
Read next because Using Cognitive Models to Improve Language Model Simulation of Human Persuasion Games overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "#351 follow-up: broader-vocab position-0 sweep at T=1.0 + position-1 suffix isolation". Matching terms: alpha, eval, rate, language, model. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17657v1 Announce Type: new Abstract: People make decisions differently in strategic interactions. Some update beliefs like a Bayesian; others exhibit biases like motivated reasoning. Although creators of large language models use simulated humans for safety evaluations and training, they often fail to cover this breadth of human behavior. We argue that cognitive science and economics provide a convenient tool for doing so, making use of mathematical models of human decision-making. We propose an approach that we call Equation-to-Behavior Prompting for guiding large language models to match cognitive models, and evaluate this approach on persuasion games based on legal decision-making. We find that large models can approximate equation-based specifications -- Bayesian updating, affine distortion, motivated updating, and Grether's $\alpha$-$\beta$ model -- using prompting, but small models fail to do so. However, training small models with reinforcement learning to adhere to mathematical rules, Equation-to-Behavior RL, reduces belief error by 26.5% in out-of-distribution parameterizations. We show that these simulations can help create diverse training environments; training small models to consider different kinds of decision-makers improves average belief change by 2.5%--12% over Bayesian-only training, even when persuading GPT-5-mini. Our work could improve human simulations for training and evaluation in increasingly realistic settings, and could also enable novel research into more complicated mathematical models of human decision-making.
Potential threat/caveat for clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)": this item discusses bias, evaluation.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17648unread
From Brewing to Resolution: Tracing the Internal Lifecycle of Code Reasoning in LLMs
Siyue Chen, Yifu Guo, Yuquan Lu, Zishan Xu, Jiaye Lin, Jianbo Lin, Siyu Zhang, Cheng Yang, Junxin Li, Yujia Li, Yu Huo, Ruixuan Wang · 2026-06-18
arXiv:2606. 17648v1 Announce Type: new Abstract: Standard accuracy metrics cannot explain why LLMs handle variable tracking but fail on semantically equivalent loops.
Read next because From Brewing to Resolution: Tracing the Internal Lifecycle of Code Reasoning in LLMs overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, text, under, eval, line, control, sweep, capability. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17648v1 Announce Type: new Abstract: Standard accuracy metrics cannot explain why LLMs handle variable tracking but fail on semantically equivalent loops. We study an internal lifecycle of code reasoning in which models first brew the answer, making it linearly recoverable many layers before it becomes self-decodable, and then diverge into one of four resolution outcomes: Resolved, Overprocessed, Misresolved, or Unresolved. Understanding this lifecycle matters because similar task accuracies can mask fundamentally different failure modes that surface-level evaluation cannot detect. We introduce a dual diagnostic framework pairing layer-wise linear probing with Context-Stripped Decoding (CSD) and apply it to six code-reasoning task families across 16 models spanning Qwen, Llama, and DeepSeek architectures. All four outcomes carry substantial mass in every task family: overall Resolved is only 41.5%, with multiple tasks below 30%. Controlled sweeps over structure, depth, and operators expose task-specific failure bottlenecks: Function Call Resolved plunges from 61.1% to 2.5% as call depth increases from one to three. Across architectures and scales, the brewing scaffold remains stable, with normalized brewing duration 24-42% across all 16 models, while resolution success varies with capability. This indicates that the scaffold is a stable empirical regularity across the tested decoder-only Transformer families, whereas resolution success covaries with capability, scale, and training. Code: https://github.com/euyis1019/llm-brewing
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses failure, evaluation.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17645unread
Beyond Domains: Reusing Web Skills via Transferable Interaction Patterns
Shiqi He, Yue Cui, Feijie Wu, Xinyu Ma, Jiaheng Lu, Yaliang Li, Bolin Ding, Mosharaf Chowdhury · 2026-06-18
arXiv:2606. 17645v1 Announce Type: new Abstract: Large language model (LLM) web agents are usually deployed as tool callers: each turn, the model reads a fresh page observation and emits one structured tool action.
Read next because Beyond Domains: Reusing Web Skills via Transferable Interaction Patterns overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)". Matching terms: completions, token, rate, compare, emit, test, completion, language. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17645v1 Announce Type: new Abstract: Large language model (LLM) web agents are usually deployed as tool callers: each turn, the model reads a fresh page observation and emits one structured tool action. When every action is a low-level primitive, horizons grow quickly and so do policy-facing LLM completions, dominating latency and cost on benchmarks such as Mind2Web and WebArena. Recent systems therefore wrap repeated interaction fragments as web skills: callable tools built from successful trajectories or induced programs, so one call can replace several primitives. However, prior skill libraries are still triggered mainly by instruction similarity or coarse site metadata, which yields low skill reuse on held-out sites and leaves much of the potential step and token reduction on the table. We present SkillMigrator, an agent that learns reusable web skills and transfers them across sites by matching layout structure rather than specific element references. Each induced skill is stored as a transferable interaction pattern (TIP): the skill paired with a structural sketch of the snapshot at induction time. At test time, SkillMigrator retrieves TIPs by layout similarity and grounds their references on the live page. The rest of the stack is standard: accessibility-snapshot observations with stable references, and fixed tool calling over primitives plus skill invocations. Compared with the state-of-the-art approaches, SkillMigrator reduces the average LLM-action count on successful trajectories by 8-10% across both WebArena and Mind2Web at matched success rate.
Potential threat/caveat for clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)": this item discusses benchmark.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17642unread
FinAcumen: Financial Multimodal Reasoning via Self-Evolving Experience Memory Harness
Pianran Guo, Pengcheng Zhou, Yucheng Jian, Shuhua Chen · 2026-06-18
arXiv:2606. 17642v1 Announce Type: new Abstract: Financial multimodal reasoning requires agents to coordinate numerical computation, retrieval, visual interpretation, and temporal grounding across heterogeneous evidence sources.
Read next because FinAcumen: Financial Multimodal Reasoning via Self-Evolving Experience Memory Harness overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, under, eval, source, rate, language, model. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17642v1 Announce Type: new Abstract: Financial multimodal reasoning requires agents to coordinate numerical computation, retrieval, visual interpretation, and temporal grounding across heterogeneous evidence sources. Existing tool-augmented agents improve execution fidelity, yet remain largely stateless across episodes, repeatedly rediscovering reasoning strategies and failure patterns. In high-stakes financial settings, this leads to unreliable tool routing, noisy retrieval, and hallucination-prone reasoning. We present FinAcumen, a financial reasoning agent framework centered on selective experience memory for tool-augmented multimodal reasoning. FinAcumen accumulates financially grounded reasoning experience from prior trajectories, distilling successful strategies and failure-derived cautionary rules into a persistent memory bank. During inference, retrieved experiences condition reasoning only when semantic relevance exceeds a calibrated threshold, while irrelevant memory is explicitly suppressed through a fallback mechanism. A deterministic financial tool environment further grounds numerical computation, retrieval, visual decoding, and answer verification.Across four financial multimodal reasoning benchmarks, FinAcumen consistently improves a frozen 8B vision-language model over finance-specialized models and approaches leading proprietary general-purpose models. Further analysis shows that selective experience activation improves reasoning reliability under retrieval uncertainty. Our code is anonymously available at https://anonymous.4open.science/r/FinAcumen
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses failure, benchmark.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17591unread
Closing the Feedback Loop: From Experience Extraction to Insight Governance in Verbal Reinforcement Learning
Yanwei Cui, Xing Zhang, Yulong Zhang, Li Shao, Xiaofeng Shi, Guanghui Wang, Peiyang He · 2026-06-18
arXiv:2606. 17591v1 Announce Type: new Abstract: Training-free verbal reinforcement learning enables LLM agents to learn from world feedback -- objective signals such as dynamic task outcomes, market returns, or demand forecasts -- by extracting verbal rules from experience and injecting them as context, updating the agent's behavior without parameter changes.
Read next because Closing the Feedback Loop: From Experience Extraction to Insight Governance in Verbal Reinforcement Learning overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, under, eval, line, extraction, without, position. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17591v1 Announce Type: new Abstract: Training-free verbal reinforcement learning enables LLM agents to learn from world feedback -- objective signals such as dynamic task outcomes, market returns, or demand forecasts -- by extracting verbal rules from experience and injecting them as context, updating the agent's behavior without parameter changes. However, in non-stationary environments these agents face a retention-forgetting dilemma: retaining stale insights causes negative transfer, while discarding them causes catastrophic forgetting when conditions recur. We identify four requirements for navigating this dilemma -- outcome-driven evaluation, persistent structured evidence, non-monotonic knowledge lifecycle, and compositional governance -- and show that existing methods invest heavily in experience extraction while underinvesting in insight governance. We propose a three-layer architecture -- rules, evidence, and skills -- connected by a feedback-driven curation loop that closes the governance gap. Rules capture distilled experience from world outcomes; evidence logs track each rule's reliability across episodes; skills govern which rules to apply, how to resolve conflicts, and when to abstain. On financial forecasting as a case study, where world feedback is naturally abundant, noisy, and non-stationary, we show that the same accumulated experience either degrades performance below the zero-shot baseline or dramatically improves accuracy and risk-adjusted returns, depending on whether the curation loop is present.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses negative, evaluation.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17577unread
Surrogate Assisted Pedestrian Protection Design via a Foundation Model Orchestrated Workflow
Osamu Ito, Akihiko Katagiri, Yoshikazu Nakagawa, Shin Saeki, Jun Shiraishi, Masato Sasaki · 2026-06-18
arXiv:2606. 17577v1 Announce Type: new Abstract: AI-driven engineering workflows face particular challenges in crash safety design: unlike aerodynamics, crash events involve highly nonlinear contact dynamics, material nonlinearity, and discrete state transitions that are difficult to capture with data-driven surrogate models.
Read next because Surrogate Assisted Pedestrian Protection Design via a Foundation Model Orchestrated Workflow overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)". Matching terms: under, eval, line, rate, trained, lora, language, model. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17577v1 Announce Type: new Abstract: AI-driven engineering workflows face particular challenges in crash safety design: unlike aerodynamics, crash events involve highly nonlinear contact dynamics, material nonlinearity, and discrete state transitions that are difficult to capture with data-driven surrogate models. To the best of our knowledge, we present the first foundation model--orchestrated workflow for crash safety design that enables surrogate-assisted exploration for pedestrian protection, reducing evaluation time from hours per CAE simulation to seconds. The workflow integrates four components: (1) a surrogate trained on CAE crash simulations to predict pedestrian leg injury metrics from design parameters, achieving an average $R^2=0.87$ and providing distribution-free conformal prediction intervals; (2) multiobjective evolutionary search (NSGA-II) to discover diverse feasible parameter sets under user-specified constraints; (3) a morphing-based geometry generator that maps parameters to topology-preserving 3D shapes; and (4) a natural-language interface in which an LLM orchestrates the workflow and a vision--language model supports semantic comparison of generated designs. In an automotive front-bumper case study, the workflow produces 35 distinct safety-compliant alternatives from a single exploration, a process that would require weeks with conventional CAE iteration. These results suggest that foundation models can serve as integration layers between ML surrogates and physics-based simulation, helping bring AI capabilities to safety-critical engineering domains.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses evaluation.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17574unread
DeepInsight: A Unified Evaluation Infrastructure Across the Physical AI Stack
Siyi Li, Chunyu Sun, Jiahao Zhang, Yuchen Kang, Wuliang Wang, Yu Qiu, Rui Jiang, Haitao Cui, Jie Chen · 2026-06-18
arXiv:2606. 17574v1 Announce Type: new Abstract: Evaluating a Physical AI stack spans operators that differ by more than three orders of magnitude -- from a single foundation-model decoding step to thousands of physics ticks of whole-body control -- varying orthogonally in modality, reward semantics, and resource profile.
Read next because DeepInsight: A Unified Evaluation Infrastructure Across the Physical AI Stack overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)". Matching terms: under, eval, source, line, rate, implement, control, full. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17574v1 Announce Type: new Abstract: Evaluating a Physical AI stack spans operators that differ by more than three orders of magnitude -- from a single foundation-model decoding step to thousands of physics ticks of whole-body control -- varying orthogonally in modality, reward semantics, and resource profile. No existing framework spans this range, so the stack is evaluated today by stitching together separate harnesses that share neither runtime nor scoring, preserving each segment's local validity but losing the shared identity needed to diagnose cross-layer regressions. We present DeepInsight, an evaluation infrastructure that serves this full spectrum on a single runtime. Rather than homogenize the regimes, it preserves their heterogeneity behind three narrow abstractions -- task, resource, and result -- each realized as one invariant shared by every subsystem: one episode driver, one resource-handle protocol implemented by every expensive backend (LLM inference and sandboxed runtimes alike), and one trace identity scheme under which every event is written. Deployed in production across all three layers of an embodied humanoid stack, this single set of invariants onboards new benchmarks largely by configuration. Where mature peer orchestrators exist -- at the foundation-model end -- it reproduces published references and peer-framework readings within their own spread, runs the same suites faster on a single node, and scales near-linearly across nodes. Its distinctive return is diagnostic: because every layer writes into one shared trace, a regression that begins in one layer and surfaces in another stays localizable on that trace -- a cross-layer payoff no federation of per-segment harnesses can reproduce.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses evaluation, benchmark.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17546unread
SEAGym: An Evaluation Environment for Self-Evolving LLM Agents
Congjie Zheng, Chuanyi Xue, Bin Liang, Jun Yang, Changshui Zhang · 2026-06-18
arXiv:2606. 17546v1 Announce Type: new Abstract: Self-evolving LLM-based agents improve mainly by changing their agent harness: the structured execution layer around a base model, including prompts, memory, tools, middleware, runtime state, and the model-tool interaction loop.
Read next because SEAGym: An Evaluation Environment for Self-Evolving LLM Agents overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)". Matching terms: under, eval, source, middle, compare, test, model. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17546v1 Announce Type: new Abstract: Self-evolving LLM-based agents improve mainly by changing their agent harness: the structured execution layer around a base model, including prompts, memory, tools, middleware, runtime state, and the model-tool interaction loop. Existing evaluations often reduce this process to isolated task scores or a single sequential curve, obscuring whether an update produces reusable improvement, overfits recent tasks, increases cost, or harms older behavior. We introduce SEAGym, an evaluation environment for measuring agent harness updates across training, validation, test, replay, and cost records. SEAGym turns Harbor-compatible benchmarks into dynamic self-evolution task sources with train batches, frozen update-validation, held-out ID and OOD transfer views, replay diagnostics, and saved snapshot and metric records. Instantiating SEAGym on Terminal-Bench 2.0 and HLE, we compare ACE, TF-GRPO, and AHE under a shared epoch/batch protocol. The results show that these evaluation views provide complementary signals about the evolution process: frequent updates may fail to improve held-out performance, useful intermediate snapshots may collapse later, and source diversity and model backend can affect harness reliability.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses evaluation, benchmark.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17507unread
LLM-as-Judge in Education: A Curriculum-Grounded Marking Pipeline
Xiwei Xu, Chen Wang, Jacky Jiang, Phil Yang, Qian Fu, Mohan Dhall, Wenjie Zhang, Liming Zhu · 2026-06-18
arXiv:2606. 17507v1 Announce Type: new Abstract: Generative AI and large language models (LLMs) are increasingly applied to question generation and automated assessment.
Read next because LLM-as-Judge in Education: A Curriculum-Grounded Marking Pipeline overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, alignment, soft, eval, line, rate, stage, language. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17507v1 Announce Type: new Abstract: Generative AI and large language models (LLMs) are increasingly applied to question generation and automated assessment. However, deploying LLMs in preparation for high-stakes exams requires more than prompt engineering; it demands software pipelines that systematically ground model outputs in authorised curriculum artefacts and marking guidelines issued by education authorities. This paper presents a curriculum-grounded, configurable LLM-as-Judge pipeline for question-level marking, co-developed with an industrial partner, to support exam preparation for university admission. The pipeline identifies the relevant topics, subtopics, and cognitive demand of a question, and assembles verifiable and authorised context to support LLM judgement. Curriculum intent is operationalised through concrete syllabus artefacts, including prescribed verbs and outcomes, performance band descriptors, glossary definitions, and marking-guideline principles. A staged LLM workflow is employed to first generate question-specific rubrics, capturing structured expectations of performance, and then derive and evaluate marking criteria used to allocate marks to student responses. This design improves consistency, transparency, and alignment with official marking practices. Preliminary evaluation shows that the proposed LLM-as-Judge pipeline delivers marking outcomes comparable to human tutors, while yielding justifications that are more traceable to authorised curriculum artefacts and marking standards. The pipeline has also been integrated into an online study platform, where early deployment data provide initial insights into operational usage and manual overrides.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17459unread
Can LLMs Be CEOs? Benchmarking Strategic Resource Reallocation with Multi-Role Agent Simulation
Yuyang Dai, Xueqing Peng, Lingfei Qian, Zhuohan Xie · 2026-06-18
arXiv:2606. 17459v1 Announce Type: new Abstract: Evaluating the decision-making capabilities of large language models (LLMs) is a growing research priority, yet existing benchmarks focus on isolated cognitive tasks such as reasoning, knowledge retrieval, and economic rationality in stylized settings.
Read next because Can LLMs Be CEOs? Benchmarking Strategic Resource Reallocation with Multi-Role Agent Simulation overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, rect, under, eval, source, line, rate, capability. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17459v1 Announce Type: new Abstract: Evaluating the decision-making capabilities of large language models (LLMs) is a growing research priority, yet existing benchmarks focus on isolated cognitive tasks such as reasoning, knowledge retrieval, and economic rationality in stylized settings. These evaluations overlook the defining challenge of real executive decision-making: integrating conflicting recommendations from specialized stakeholders under information asymmetry, organizational constraints, and temporal dependencies. We introduce \textsc{CEO-Bench}, a multi-agent benchmark that evaluates LLMs on CEO-level strategic resource reallocation -- the process of redirecting capital across business units in a multi-round, constraint-rich organizational environment. In \textsc{CEO-Bench}, LLM agents receive conflicting advice from four role-conditioned C-suite advisors (CFO, CTO, COO, CMO), each with private signals and distinct priorities, and must synthesize these into a concrete allocation plan evaluated along four dimensions: role integration, conditional boldness, history-sensitive judgment, and plan validity. Experiments across five frontier models on 13 scenarios reveal that all models achieve high structural validity but diverge sharply on strategic calibration -- the hardest capability layer. We identify systematic failure modes including single-advisor capture, conservative default under ambiguity, and historical amnesia, and uncover a structural integration-boldness tradeoff: models that engage more deeply with conflicting perspectives tend to produce less decisive action. These findings delineate the current capability boundary of LLMs as organizational decision-makers and inform the design of future AI-assisted executive systems.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses failure, evaluation, benchmark.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17454unread
Dissecting model behavior through agent trajectories
Gaurav Gupta, Vatshank Chaturvedi, Jun Huan, Anoop Deoras · 2026-06-18
arXiv:2606. 17454v2 Announce Type: new Abstract: AI agent performance is not just a modeling problem, it is fundamentally a systems problem.
Read next because Dissecting model behavior through agent trajectories overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, text, latin, alignment, rate, full, stage, test. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17454v2 Announce Type: new Abstract: AI agent performance is not just a modeling problem, it is fundamentally a systems problem. The advanced capabilities of models are realized through agent harnesses. Therefore, a gap between model assumptions and harness behavior can easily prevent the model's full capabilities from translating into agent performance. We formalize this as the `intent-execution' gap: the mismatch between what the model intends and what the harness executes, and vice versa. We argue that minimizing this intent-execution gap is as important as other aspects of harness design such as tools and execution loops. To illustrate the impact of this harness-model alignment, we develop a simple and customizable harness called `Simple Strands Agent' (SSA). SSA aims to find the bulk of common patterns which generalize across different model families (such as Claude, Gemini, GPT, Grok, Qwen), as well as a small number of model-specific preferences. We make two contributions: (i) we reproduce or improve on the pass@1 performance reported by diverse model-provider families on popular agentic benchmarks (SWE-Pro, SWE-Verified and Terminal-Bench-2), and (ii) building on an analysis of 138k trajectories generated by SSA, we look beyond the pass@1 numbers which tend to be relatively even across frontier models. By representing agent trajectories in code state-spaces, we observe model-level differences in problem-solving behavior. Finer-grained metrics such as edit frequency, testing activity, and phase-transitions reveal how individual models allocate effort across different stages of autonomous problem solving.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses benchmark.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17453unread
MapSatisfyBench: Benchmarking Satisfaction-Aware Map Agents through Behavior-Grounded Implicit Decision Factors
Lubin Bai, Mengyu Cao, Sixue Wang, Zhongwei Wan, Yue Pan, Jiale Hou, Xiang Li, Xiuyuan Zhang · 2026-06-18
arXiv:2606. 17453v2 Announce Type: new Abstract: Large language model agents are increasingly integrated into map services.
Read next because MapSatisfyBench: Benchmarking Satisfaction-Aware Map Agents through Behavior-Grounded Implicit Decision Factors overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)". Matching terms: under, eval, source, rate, full, chain, factor, completion. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17453v2 Announce Type: new Abstract: Large language model agents are increasingly integrated into map services. Since map services are embedded in everyday-life scenarios rather than professional task settings, users often express their needs informally, resulting in underspecified queries with many unspoken needs, namely, implicit decision factors that are critical for user satisfaction. Although clarification is an effective way to mitigate this issue, it increases user burden in daily interaction, and a capable agent should first proactively recover such factors from available information sources. However, evaluating this ability is challenging. The first challenge is to determine which implicit decision factors are suitable for evaluation. A factor is evaluable only if it affects user acceptance and can be recovered from information available to the agent before it responds. Second, user satisfaction cannot be reliably represented by a single reference answer, requiring a benchmark that converts satisfaction-relevant factors into objective and quantifiable evaluation targets. To address these challenges, we propose a restore-identify-filter framework that reconstructs complete user needs from behavior-chain evidence, identifies implicit decision factors, and retains only those supported by pre-query evidence. Building on this methodology, we construct MapSatisfyBench from large-scale, real-world anonymized user data and annotate ground truth from five dimensions and enables full-chain evaluation of satisfaction-aware map agents. Experiments show that current agents generally perform well on explicit task completion, but remain limited in satisfying implicit decision factors and proactively acquiring the evidence needed for satisfaction-aware decisions. These findings establish MapSatisfyBench as a benchmark for shifting map-agent evaluation from task completion toward satisfaction-aware spatial decision making.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses evaluation, benchmark.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17450unread
A Machine-Learned Comorbidity Index
Suleman Baloch, Kishlay Jha, Alberto M. Segre, Philip M. Polgreen, Bijaya Adhikari · 2026-06-18
arXiv:2606. 17450v1 Announce Type: new Abstract: Traditional comorbidity scores (e.
Read next because A Machine-Learned Comorbidity Index overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, strong, eval, line. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17450v1 Announce Type: new Abstract: Traditional comorbidity scores (e.g., Charlson and Elixhauser) are widely used for risk adjustment and patient stratification, but they have two key limitations: (i) they are largely mortality-centric and do not align well with other clinical outcomes, and (ii) their linear, rule-based structure cannot capture nonlinear, outcome-specific risk relationships. We propose a Machine-Learned Comorbidity Index (MLCI) that maps diagnosis codes to a single scalar by maximizing the normalized Hilbert-Schmidt Independence Criterion (nHSIC) between the learned score and multiple clinical outcomes. MLCI captures nonlinear risk-outcome dependence and is supported by a theory that characterizes when a unified, informative admission-level ordering can be achieved across outcomes. Empirical results on multiple benchmark electronic health record (EHR) datasets show that MLCI outperforms strong baselines across multiple evaluation metrics.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses limitation, limitations, evaluation, benchmark.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17443unread
Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems
Xi Chu, Yupeng Hou · 2026-06-18
arXiv:2606. 17443v1 Announce Type: new Abstract: Large language models (LLMs) are becoming a major way for consumers to find products, but we do not yet understand how brands compete in this new channel.
Read next because Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)". Matching terms: under, good, rate, test, language, model. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17443v1 Announce Type: new Abstract: Large language models (LLMs) are becoming a major way for consumers to find products, but we do not yet understand how brands compete in this new channel. We study brand dynamics in LLM recommendations using skincare products -- a category where consumers cannot easily judge quality before buying and must rely on brand reputation -- across three commercial LLMs (GPT-4o-mini, Claude Sonnet, Gemini 3 Flash), with a robustness check on search goods. In three experiments, we find: (1) a Conditional Monopoly where well-known brands get recommended 100% of the time (IAI = 10.0) when all products have the same specifications, but this dominance disappears with less than a +0.1-star rating advantage for a competitor; (2) authority-style marketing language, including fabricated clinical-evidence claims, breaks this monopoly at a Bias Surplus Value equal to +0.17 rating points, with each model responding differently; and (3) a social dilemma in multi-brand GEO competition: when all brands adopt the same optimization strategy, individual payoff falls from +0.802 to +0.007 in our payoff proxy, and non-participating brands receive zero recommendations in our tests. Our results suggest that generative engine optimization (GEO) should be studied not only as a security risk, but also as an emerging marketing practice that shapes market competition.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses bias, robustness.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17339unread
SpeechDx: A Multi-Task Benchmark for Clinical Speech AI
Sejal Bhalla, Larry Kieu, Aina Merchant, Eyal de Lara, Alex Mariakakis · 2026-06-18
arXiv:2606. 17339v1 Announce Type: new Abstract: Speech offers a uniquely informative window into health by simultaneously engaging neurological, motor, respiratory, and vocal systems.
Read next because SpeechDx: A Multi-Task Benchmark for Clinical Speech AI overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, strong, under, eval, line, compare, stage, test. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17339v1 Announce Type: new Abstract: Speech offers a uniquely informative window into health by simultaneously engaging neurological, motor, respiratory, and vocal systems. Current clinical speech AI methods have largely progressed through isolated condition-specific studies, making results difficult to compare and generalization difficult to assess. We introduce SpeechDx, a large-scale benchmark for clinical speech AI spanning 12 datasets and 27 tasks across diverse health conditions. To enable evaluation across shared clinical mechanisms, SpeechDx structures tasks by the stage of speech production they disrupt: conceptualization, formulation, and articulation. The benchmark tests generalization by including tasks with limited labeled data and evaluating the same health condition across multiple datasets, distinguishing clinically meaningful patterns from dataset artefacts. We systematically evaluate 12 state-of-the-art audio encoders across all tasks and under zero-shot cross-condition transfer. Results show that large-scale speech models represent the strongest overall baselines, domain-specific models improve performance only on closely matched tasks, and no current representation generalizes reliably across the clinical speech landscape. SpeechDx establishes a shared evaluation framework for tracking progress toward general-purpose clinical speech representations
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation, benchmark.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17328unread
MemTrace: Probing What Final Accuracy Misses in Long-Term Memory
Xianxuan Long, Zhikai Chen, Shenglai Zeng, Shouren Wang, Kai Guo, Jiliang Tang · 2026-06-18
arXiv:2606. 17328v1 Announce Type: new Abstract: LLM agents increasingly maintain long-term memory of user facts across sessions.
Read next because MemTrace: Probing What Final Accuracy Misses in Long-Term Memory overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: rect, correct, eval, control, does. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17328v1 Announce Type: new Abstract: LLM agents increasingly maintain long-term memory of user facts across sessions. Yet such memory is usually evaluated by aggregating accuracy over question rows or episodes. Because this approach scores question rows independently, even when several questions probe the same fact, it cannot show how that fact behaves as conditions change. We introduce MemTrace, a benchmark whose unit of measurement is the knowledge point: a single typed fact about the user, rather than an individual question. MemTrace probes each fact along three controlled dimensions: memory age, defined by how many sessions ago the fact appeared in the history; question type, covering current state, earlier state, and trajectory of change; and evidence condition, covering present, missing, and contradicted-by-false-premise settings. Evaluating 13 memory-system configurations across four paradigms, we find that similar pooled accuracy hides different failures: recovering a fact's current and earlier states does not imply tracking how it changed, and safe abstention does not imply correcting a false premise. The dominant bottleneck is evidence use, not retrieval: when systems fail, the evidence was retrievable 10 times more often than it was missing. These results suggest that improving long-term memory requires better use of reachable evidence, not simply more storage or retrieval.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses failure, failures, benchmark.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17312unread
Quantifying Consistency in LLM Logical Reasoning via Structural Uncertainty
Baishali Chaudhury, Mengdie Flora Wang, Hyunji Hayley Park, Rahul Ghosh, Sungmin Hong, Jae Oh Woo · 2026-06-18
arXiv:2606. 17312v1 Announce Type: new Abstract: Large language models can arrive at the same answer through reasoning paths that are unstable, contradictory, or difficult to rank consistently -- a failure mode especially prevalent in multi-step deductive reasoning.
Read next because Quantifying Consistency in LLM Logical Reasoning via Structural Uncertainty overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: rect, under, correct, eval, rate, candidates, candidate, language. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17312v1 Announce Type: new Abstract: Large language models can arrive at the same answer through reasoning paths that are unstable, contradictory, or difficult to rank consistently -- a failure mode especially prevalent in multi-step deductive reasoning. Existing methods assess reliability primarily through output dispersion -- measuring how much sampled answers differ -- but this discards a complementary signal: whether the model can consistently rank competing reasoning candidates. We propose structural uncertainty, a consistency-aware framework derived from the stability of self-preference-induced rankings over sampled reasoning solutions. Given a query, we generate multiple candidate solutions and ask the model to judge pairwise preferences among its own outputs. We aggregate self-preferences into ranking distributions via Bradley-Terry modeling with PageRank, and decompose the signal into two entropy-based components: across-trial ranking instability and within-trial candidate ambiguity. Across five LLMs and eight benchmarks, structural signals provide information complementary to answer dispersion: on logical and mathematical reasoning tasks, the combination improves identification of unreliable instances, while on factual retrieval the structural signal collapses toward uniformity, diagnosing a regime boundary where reasoning-level consistency evaluation is uninformative. The two components relate differently to accuracy: within-trial ambiguity correlates positively with correctness -- consistent with settings where multiple plausible solution paths remain competitive -- while across-trial instability correlates negatively, signaling unreliable reasoning. Structural uncertainty is best understood not as a universal confidence estimator, but as a regime-sensitive evaluator of logical reasoning consistency.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses failure, negative, evaluation, benchmark.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17269unread
Skill-Constrained Model Predictive Control for Resilient Manufacturing Supply Chains
Carlos Eduardo Sanoja · 2026-06-18
arXiv:2606. 17269v1 Announce Type: new Abstract: In skill-constrained production-inventory systems, the qualified human capacity available tomorrow depends on training decisions made today: production requires certified workers, certifications decay unless maintained, and training consumes the same scarce worker hours that production needs now.
Read next because Skill-Constrained Model Predictive Control for Resilient Manufacturing Supply Chains overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, class, under, eval, rate, control, chain, trained. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17269v1 Announce Type: new Abstract: In skill-constrained production-inventory systems, the qualified human capacity available tomorrow depends on training decisions made today: production requires certified workers, certifications decay unless maintained, and training consumes the same scarce worker hours that production needs now. We study a closed-loop skill-constrained model predictive controller that, at every shift, solves a finite-horizon mixed-integer program over production, inventory, backlog, and training, with binary predicted certification, hard production eligibility, and an interpretable terminal value that prices certified-capacity gaps at the horizon boundary; only the first-period action is applied before replanning. On synthetic, seed-controlled SkillChain-Gym scenarios - announced and surprise new-skill shocks, demand shocks, absenteeism, forecast- and availability-quality modes, capacity-boundary and training-rate sweeps, and negative controls - we evaluate the controller against production-only and maintenance-only ablations, static cross-training insurance plans, and a strong reactive heuristic, under an ex-ante locked configuration and paired statistics. The result is regime dependence, not superiority: no policy class dominates. Predictive control helps when skill or labor bottlenecks are forecastable early enough for training to complete; lean static insurance remains hard to beat under surprise shocks, near the demand-capacity boundary, and wherever pre-shock slack makes insurance cheap. Attribution ablations separate certification maintenance, re-acquisition of lapsed certifications, and greenfield skill acquisition. Forecastability, not adaptivity per se, decides when predictive control pays.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses negative.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17266unread
SkillChain-Gym: A Benchmark for Reskilling-Aware Production-Inventory Control under Disruptions
Carlos Eduardo Sanoja · 2026-06-18
arXiv:2606. 17266v1 Announce Type: new Abstract: Production planning increasingly has to treat workforce capability as a decision variable: certifications lapse when skills are not maintained, new products require skills the current workforce does not hold, and reskilling competes for the same worker hours needed for production.
Read next because SkillChain-Gym: A Benchmark for Reskilling-Aware Production-Inventory Control under Disruptions overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, strong, fill, class, under, eval, line, rate. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17266v1 Announce Type: new Abstract: Production planning increasingly has to treat workforce capability as a decision variable: certifications lapse when skills are not maintained, new products require skills the current workforce does not hold, and reskilling competes for the same worker hours needed for production. Existing operations benchmarks usually treat labor as exogenous, while workforce-planning models with skills and learning are rarely released as reusable testbeds. We introduce SkillChain-Gym, a benchmark specification for reskilling-aware production-inventory control: a single-site environment with stylized worker skill-state dynamics, hard threshold certification, forgetting, and capacity-consuming training actions constrained by the same per-worker time budget as production. The benchmark includes seed-controlled disruption scenarios, three feasibility modes with projection diagnostics, deterministic replay, and metrics covering operations, resilience, capability growth, and training-access distribution. We evaluate production-only, reactive adaptive, water-filling adaptive, and static-insurance policies with budget variants over 60-shift horizons with paired statistical tests. The results are regime-dependent rather than a ranking. Training-capable policies dominate the production-only baseline, and maintenance training is necessary under forgetting even without disruptions. Among training-capable classes, adaptive training helps when bottlenecks are visible in the forecast, while a lean static cross-training plan, a deliberately favorable comparator whose structure encodes relevant skill contingencies, acts as strong insurance under surprise shocks and absenteeism. Capacity slack and the forgetting rate govern the boundary between these regimes. No policy class dominates across regimes, motivating forecast-driven controllers that decide when to buy skill insurance and when to react.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses benchmark.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17220unread
When Rules Learn: A Self-Evolving Agent for Legal Case Retrieval
Mingxu Tao, Jiawei Hu, Xian Zhou, Wenpeng Hu, Jiajun Cheng, Yunbo Cao, Zhunchen Luo, Guotong Geng · 2026-06-18
arXiv:2606. 17220v1 Announce Type: new Abstract: Legal case retrieval remains challenging due to the complexity of legal language and the need for precise lexical alignment between queries and relevant cases.
Read next because When Rules Learn: A Self-Evolving Agent for Legal Case Retrieval overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, under, alignment, eval, line, rate, without, language. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17220v1 Announce Type: new Abstract: Legal case retrieval remains challenging due to the complexity of legal language and the need for precise lexical alignment between queries and relevant cases. Although dense retrieval models have achieved notable progress, empirical studies show that BM25 continues to serve as a strong baseline in this domain. It motivates us to propose a self-evolving framework for rule-driven query rewriting that enhances BM25 without any parameter training. The framework equips an LLM-based agent with an automatic evaluation environment, enabling it to iteratively create rewriting rules, plan validation experiments over rule combinations, and eliminate ineffective rules based on historical feedbacks. We evaluate our method on the Chinese legal case retrieval benchmark LeCaRD-v2. Experimental results demonstrate that the proposed framework outperforms non-evolutionary baselines, including human-designed rules and greedy rule selection, particularly when powered by a highcapacity core LLM. We also conduct detailed analyses to investigate the mechanisms underlying self-evolution. Our findings reveal that LLM's capabilities to leverage previous experimental results and its intrinsic knowledge of rule elimination play critical roles in refining the rule set via self-evolution.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation, benchmark.
source ↗open
score 100arxiv cs.AI (Artificial Intelligence)arxiv:2606.17209unread
Beyond Parallel Sampling: Diverse Query Initialization for Agentic Search
Sidhaarth Murali, Jo\~ao Coelho, Jingjie Ning, Jo\~ao Magalh\~aes, Bruno Martins, Chenyan Xiong · 2026-06-18
arXiv:2606. 17209v1 Announce Type: new Abstract: Test-time scaling for agentic search typically increases depth (i.
Read next because Beyond Parallel Sampling: Diverse Query Initialization for Agentic Search overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, eval, token, candidates, candidate, test, model. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2606.17209v1 Announce Type: new Abstract: Test-time scaling for agentic search typically increases depth (i.e., more turns and tokens per trajectory) or breadth (i.e., more parallel rollouts). Here we focus on breadth scaling, showing that standard parallel sampling yields diminishing returns, tracing this to query redundancy at the first turn. When models issue similar first queries across rollouts, the threads retrieve overlapping evidence, and subsequent turns are conditioned on this shared retrieval. We address this limitation with DivInit, a training-free intervention at the first turn. Rather than sampling k independent first queries, DivInit draws n candidates from a single call, picks k < n diverse seeds, and runs them as parallel trajectories. Across five open-weight models and eight benchmarks, DivInit consistently improves over standard parallel sampling, with average gains of five to seven points on multi-hop QA at matched compute. Code available at https://github.com/cxcscmu/diverse-query-initialization
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses limitation, benchmark.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15396unread
CHILLGuard: Towards Fine-Grained Chinese LLM Safety Guardrail with Scalable Data Construction and Model-aware Preference Alignment
Wenbo Yu, Bohua Wang, Hao Fang, Kuofeng Gao, Jingru Zeng, Xiaochen Yang, Tianyi Zhang, Xiaoxiao Ma, Jiawei Kong, Hao Wu, Bin Chen, Shu-Tao Xia, Min Zhang · 2026-06-17
arXiv:2606. 15396v1 Announce Type: new Abstract: Malicious content generated from large language models (LLMs) could pose severe safety risks and ethical concerns.
Read next because CHILLGuard: Towards Fine-Grained Chinese LLM Safety Guardrail with Scalable Data Construction and Model-aware Preference Alignment overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, class, rect, under, alignment, eval, source, line. Source: arxiv cs.CL (NLP).
arXiv:2606.15396v1 Announce Type: new Abstract: Malicious content generated from large language models (LLMs) could pose severe safety risks and ethical concerns. While existing LLM safety guardrails excel in English or multilingual settings, they lack adaptation to Chinese-specific regulatory policies, cultural context and linguistic nuances, failing to support fine-grained risk classification for diverse deployment needs. In this paper, we introduce a 5-macro, 31-micro category fine-grained risk taxonomy for Chinese scenarios, and build CHILLGuard: a dedicated Chinese LLM content safety guardrail. To address the critical scarcity of high-quality annotated Chinese safety data, we propose a scalable multi-stage data construction pipeline: we expand multi-source corpus via retrieval-augmented generation, generate implicit harmful samples through prompt engineering rewriting, and refine high-quality data via multi-model voting-based label calibration. Based on this, we build CHILLGuardTrain, a large-scale training set with 405,007 samples, and CHILLGuardTest, a rigorously curated annotated test set with 51,745 samples. We then train CHILLGuard on CHILLGuardTrain under a generator-classifier collaborative framework via Model-aware Direct Preference Optimization. Extensive experiments under multiple settings demonstrate the state-of-the-art performance of CHILLGuard, e.g., a 15.92% improvement of F1 score over Qwen3Guard-8B-Strict on our benchmark. We will release our resources at https://github.com/cswbyu/CHILLGuard.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses benchmark.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15390unread
Not All Skills Help: Measuring and Repairing Agent Knowledge
Yixuan Wang, Yiyang Zhou, Yiming Liang, Congyu Zhang, Fuxiao Liu, Jiawei Zhou, Huaxiu Yao · 2026-06-17
arXiv:2606. 15390v1 Announce Type: new Abstract: LLM agents can improve without weight updates by accumulating natural-language skills from experience, but current systems entrust every decision about which skills to keep and how to apply them to LLM judgment alone.
Read next because Not All Skills Help: Measuring and Repairing Agent Knowledge overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, latin, line, rate, without, alone, test, completion. Source: arxiv cs.CL (NLP).
arXiv:2606.15390v1 Announce Type: new Abstract: LLM agents can improve without weight updates by accumulating natural-language skills from experience, but current systems entrust every decision about which skills to keep and how to apply them to LLM judgment alone. We argue that this conflates two distinct roles: generating a skill from experience is a creative act that judgment handles well, while deciding whether that skill actually helps requires empirical evidence across many tasks. Measuring per-skill causal contributions via randomized masking, we find that skill libraries exhibit pervasive causal heterogeneity: individual skills routinely help on some task types while hurting on others, yet their opposing effects cancel in aggregate, making them invisible to global curation methods. We propose ASSAY, a framework that separates generation from curation: it computes a per-skill causal attribution on a small development set, restructures the library offline, and suppresses skills with negative predicted effect for each test task. Across seven base models spanning four providers and two benchmarks (AppWorld and tau-bench), ASSAY consistently improves over prior skill-curation approaches. On AppWorld's hardest split, DeepSeek-V3 achieves 69.3% task-goal completion (47.4% relative improvement), a new state of the art among all published methods including weight-tuned approaches. On tau-bench retail, GPT-4.1 improves by 8.7% relative, advancing past o4-mini, o1, and GPT-4.5 on the public leaderboard without any weight modification. Ablation traces the dominant gain to per-task masking, confirming that the bottleneck is matching skills to tasks at inference time, not removing bad skills globally. Code is available at https://github.com/aiming-lab/assay.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses negative, benchmark.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15345unread
Beyond Monolingual Deep Research: Evaluating Agents and Retrievers with Cross-Lingual BrowseComp-Plus
Yuheng Lu, Qingcheng Zeng, Heli Qi, Puxuan Yu, Fuheng Zhao, Rui Yang, Hitomi Yanaka, Naoto Yokoya, Weihao Xuan · 2026-06-17
arXiv:2606. 15345v1 Announce Type: new Abstract: Deep research agents are increasingly evaluated on their ability to search for evidence, reason over retrieved sources, and produce grounded answers.
Read next because Beyond Monolingual Deep Research: Evaluating Agents and Retrievers with Cross-Lingual BrowseComp-Plus overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, rect, eval, source, rate, control, full, another. Source: arxiv cs.CL (NLP).
arXiv:2606.15345v1 Announce Type: new Abstract: Deep research agents are increasingly evaluated on their ability to search for evidence, reason over retrieved sources, and produce grounded answers. Existing browsing benchmarks, however, largely assume that the user's query and the supporting evidence are written in the same language, leaving open whether agentic search systems can operate when relevant evidence appears in another language. We introduce XBCP (Cross-lingual BrowseComp-Plus), a controlled benchmark that preserves the English question-and-answer space of BrowseComp-Plus but varies the languages of the supporting documents. XBCP instantiates two complementary settings: in the cross-lingual setting, each query is paired with evidence in a single assigned language. In the multilingual setting, the full evidence corpus is distributed equally and randomly across 12 languages spanning high-resource and low-resource regimes. We evaluate four deep research agents using sparse and dense multilingual retrievers, measuring answer accuracy, evidence recall, search behavior, calibration, citation fidelity, and oracle retrieval. Results reveal substantial degradation when evidence is translated. Even strong, dense retrievers lose evidence recall, and agents become less calibrated and cite evidence less reliably. Notably, accuracy remains lower even when all gold evidence is supplied directly. These findings suggest that cross-lingual deep research exposes both retrieval failures and an independent, agent-side difficulty in integrating language-mismatched evidence.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses failure, failures, benchmark.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15335unread
Privacy-Preserving Text Sanitization for Distributed Agents Collaboration via Disentangled Representations
Xuan Liu, Hefeng Zhou, Sicheng Chen, Chao Yang, Xingcheng Xu, Jingjing Qu, Jiong Lou, Jie LI, Xia Hu · 2026-06-17
arXiv:2606. 15335v1 Announce Type: new Abstract: When distributed agents exchange text across organizational boundaries, privacy leakage arises not only from explicit identifiers but also from distributional signatures such as formatting conventions, vocabulary choices, and syntactic patterns.
Read next because Privacy-Preserving Text Sanitization for Distributed Agents Collaboration via Disentangled Representations overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, text, under, alignment, distributional, source, token, rate. Source: arxiv cs.CL (NLP).
arXiv:2606.15335v1 Announce Type: new Abstract: When distributed agents exchange text across organizational boundaries, privacy leakage arises not only from explicit identifiers but also from distributional signatures such as formatting conventions, vocabulary choices, and syntactic patterns. We propose DiSan(Disentangled Sanitization), a privacy-preserving sanitization framework and a built-in component of Intern-Shannon for multi-agent collaboration. DiSan uses a two-stream encoder to factorize text into a source-invariant role subspace that preserves task semantics and a source-identifying style subspace that remains local. Federated proto-type alignment and adversarial regularization enable joint training without centralizing raw text. Experiments show that identifier-level masking is insufficient: masking 19.2% of tokens reduces TF-IDF stylometric attribution by only 18.6%. By contrast, DiSan reduces answer-level PII exposure by 20 times while maintaining 83% answer faithfulness on a distributed multi-agent RAG benchmark, and lowers Enron stylometric attribution by 73.2% under TF-IDF and 70.6% under a neural probe.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses adversarial, benchmark.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15333unread
Replay What Matters: Off-Policy Replay for Efficient LLM Reinforcement Unlearning
Zirui Pang, Chenlong Zhang, Haosheng Tan, Zhuoran Jin, Jiaheng Wei, Zixin Zhong · 2026-06-17
arXiv:2606. 15333v1 Announce Type: new Abstract: LLM unlearning has emerged as a cost-effective alternative to full retraining for removing hazardous knowledge from pretrained models while preserving general utility.
Read next because Replay What Matters: Off-Policy Replay for Efficient LLM Reinforcement Unlearning overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: rect, full, trained, on-policy, stage, model. Source: arxiv cs.CL (NLP).
arXiv:2606.15333v1 Announce Type: new Abstract: LLM unlearning has emerged as a cost-effective alternative to full retraining for removing hazardous knowledge from pretrained models while preserving general utility. Recent RL-based methods such as RULE reformulate unlearning as learning a refusal behavior, but their on-policy optimization repeatedly samples from the same forget and retain/boundary prompts throughout training. We identify a critical inefficiency in this process: easy cases quickly converge and provide little useful gradient signal, while hard cases near the forget/retain boundary continue to produce low-reward rollouts that are discarded after a single use. To address this issue, we propose ReRULE, an off-policy replay enhancement for reinforcement unlearning. ReRULE stores low-reward hard-case rollout groups in a replay buffer during early GRPO training and reuses them in later stages through importance-sampled off-policy updates, redirecting computation toward boundary cases that still require learning. Theoretically, we show that ReRULE yields a tighter hard-case convergence bound than pure on-policy RULE. Empirically, ReRULE improves MUSE-Books Retain Quality from 46.3 to 56.2 while adding only 5--11% training time across benchmarks. Its limited improvement on the simpler TOFU setting further supports the intended conditional behavior: replay is most beneficial when the hard/easy disparity is pronounced.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses benchmark.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15307unread
Adapting Reinforcement Learning with Chain-of-Thought Supervision for Explainable Detection of Hateful and Propagandistic Memes
Mohamed Bayan Kmainasi, Mucahid Kutlu, Ali Ezzat Shahroor, Abul Hasnat, Firoj Alam · 2026-06-17
arXiv:2606. 15307v1 Announce Type: new Abstract: Hateful and propagandistic memes exploit the interplay between images and text to convey harmful intent that neither modality reveals alone.
Read next because Adapting Reinforcement Learning with Chain-of-Thought Supervision for Explainable Detection of Hateful and Propagandistic Memes overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, strong, text, class, under, eval, source, line. Source: arxiv cs.CL (NLP).
arXiv:2606.15307v1 Announce Type: new Abstract: Hateful and propagandistic memes exploit the interplay between images and text to convey harmful intent that neither modality reveals alone. Although thinking-based multimodal large language models (MLLMs) have advanced vision-language understanding, their application to meme content moderation remains underexplored. We propose a reinforcement learning-based post-training method that improves classification performance and reference-based explanation quality in thinking-based MLLMs via task-specific rewards and Group Relative Policy Optimization (GRPO). Concretely, we (i) conduct a systematic empirical study of off-the-shelf MLLMs for hateful and propagandistic meme understanding across English and Arabic benchmarks, (ii) extend existing meme datasets with weakly supervised chain-of-thought (CoT) rationales via distillation and multi-LLM fine-grained propaganda annotations, (iii) introduce a GRPO-based objective with thinking-length regularization that jointly optimizes classification accuracy and explanation quality, and (iv) investigate self-supervised GRPO on unlabeled memes using consensus-based pseudo-labels. Experiments on the Hateful Memes and ArMeme benchmarks show that our approach improves over previously reported results on FHM accuracy (up to +2.1%, from 79.9% to 82.0%) and on ArMeme macro-F1 (up to +7.6 points, from 0.536 to 0.612 with explanations; +6.1 compared to the original ArMeme benchmark), while also generating natural-language explanations. On ArMeme, sequence-classification baselines remain stronger in terms of raw accuracy, whereas our approach provides more balanced per-class performance along with explanations. We publicly release our code, data extensions, and evaluation resources.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation, benchmark.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15266unread
Evaluating and Preserving Lexical Stress in English-to-Chinese Speech-to-Speech Translation
Yuchen Song, Xi Chen, Mingze Li, Satoshi Nakamura · 2026-06-17
arXiv:2606. 15266v1 Announce Type: new Abstract: Speech-to-speech translation (S2ST) systems have achieved impressive progress in semantic accuracy and speech naturalness.
Read next because Evaluating and Preserving Lexical Stress in English-to-Chinese Speech-to-Speech Translation overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, under, eval, rate, capability, language. Source: arxiv cs.CL (NLP).
arXiv:2606.15266v1 Announce Type: new Abstract: Speech-to-speech translation (S2ST) systems have achieved impressive progress in semantic accuracy and speech naturalness. However, the cross-lingual transfer of lexical stress, a vital cue for emphasis and speaker intent, remains heavily underexplored, compounded by a lack of reliable automatic evaluation metrics for tonal languages like Chinese. We investigate English-to-Chinese S2ST stress transfer by constructing a stress-annotated Chinese dataset and an XLS-R-based Mandarin stress detector. Integrating this with the English EmphAssess system, we propose a novel objective metric for cross-lingual stress evaluation. Furthermore, we fine-tune CosyVoice3 to build a stress-aware S2ST system. Experiments demonstrate that our proposed S2ST architecture significantly outperforms existing systems in stress translation capability while maintaining competitive translation quality. Furthermore, our evaluation metric exhibits a strong correlation with human subjective judgments.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15191unread
AmchiBias: Measuring Stereotypical Bias in Goan Identity Groups with a Minimal Pair Dataset in English and Konkani
Michelle Barbosa, Sebastian Pad\'o, Franziska Weeber · 2026-06-17
arXiv:2606. 15191v1 Announce Type: new Abstract: Socio-cultural stereotypical bias is an important consideration in the development and deployment of NLP systems.
Read next because AmchiBias: Measuring Stereotypical Bias in Goan Identity Groups with a Minimal Pair Dataset in English and Konkani overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, strong, eval, source, language, model. Source: arxiv cs.CL (NLP).
arXiv:2606.15191v1 Announce Type: new Abstract: Socio-cultural stereotypical bias is an important consideration in the development and deployment of NLP systems. It is however often considered only at the national level, despite rich subnational socio-cultural structures. We present AmchiBias, the first benchmark for measuring socio-cultural stereotypical bias for the Indian state of Goa with its unique historically multicultural setting. It covers various Goan identity groups and comprises 313 minimal pairs across eight sociodemographic dimensions in both English and Devanagari Konkani. We then evaluate stereotypical bias in five multilingual encoder models on this benchmark. We find near-chance scores in Konkani, reflecting language incompetence for general multilingual models and a lack of Goan cultural competence for Indian language models. Queried in English, models with a stronger Indian language coverage show higher bias for pan-Indian groups than hyperlocal Goan groups. This suggests the English signal reflects pan-Indian pretraining associations rather than genuine Goan cultural knowledge. Our findings highlight a critical gap in low-resource multilingual NLP evaluation for hyperlocal community identities.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses bias, evaluation, benchmark.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15152unread
Can Agents Read the Room? Benchmarking Visual Social Intelligence in Multimodal Simulation
Shijun Wan, Xuehai Wu, Jiwen Zhang, Siyuan Wang, Zhongyu Wei · 2026-06-17
arXiv:2606. 15152v1 Announce Type: new Abstract: Social interaction depends on both language and visible social signals, such as facial expressions, posture, gaze, and emotional shifts.
Read next because Can Agents Read the Room? Benchmarking Visual Social Intelligence in Multimodal Simulation overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, text, rect, under, eval, test, language. Source: arxiv cs.CL (NLP).
arXiv:2606.15152v1 Announce Type: new Abstract: Social interaction depends on both language and visible social signals, such as facial expressions, posture, gaze, and emotional shifts. Yet existing social-agent benchmarks are largely text-based and rarely test whether multimodal agents can use visual cues to guide interaction. We introduce \textsc{\benchmarkname{}}, a benchmark evaluating visual social intelligence in multimodal social simulation. It contains 240 scenarios, 585 role instances, and 2,340 role-task instances, combining aligned textual-visual evidence, structured role profiles, and four role-level tasks: expression task, characteristic task, interaction regulation task, and interaction outcome task. Evaluating seven recent MLLMs under verbalized-vision and direct-vision reveals a clear gap between local role enactment and interaction management: role-specific expression and conflict handling are near saturation, whereas interaction regulation and visually grounded outcome achievement remain substantially more difficult. The code is released at https://github.com/JunsWan/AgentViSS, and the dataset is available at https://huggingface.co/datasets/JunsWan/AgentViSS.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses benchmark.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15144unread
PACUTE: Phonology-, Affix-, and Character-level Understanding of Tokens for Filipino
Jann Railey Montalan, David Demitri Africa, Jimson Paulo Layacan, Richell Isaiah Flores, Ivan Yuri De Leon, Lance Calvin Gamboa · 2026-06-17
arXiv:2606. 15144v1 Announce Type: new Abstract: Large language models (LLMs) process text as sequences of subword tokens, which can obscure the character-level and morphological structure that underlies word formation.
Read next because PACUTE: Phonology-, Affix-, and Character-level Understanding of Tokens for Filipino overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, word, under, eval, token, alone, position, language. Source: arxiv cs.CL (NLP).
arXiv:2606.15144v1 Announce Type: new Abstract: Large language models (LLMs) process text as sequences of subword tokens, which can obscure the character-level and morphological structure that underlies word formation. This limitation is most acute for languages with non-concatenative morphology, where standard tokenizers systematically misalign token boundaries with morpheme boundaries. We introduce PACUTE, a diagnostic benchmark of 4,600 tasks designed to evaluate morphological understanding in Filipino, a language characterized by productive infixation, reduplication, and diacritic-driven lexical distinctions that are typically absent from written text. PACUTE includes a hierarchical diagnostic framework of six compositional levels that localizes where morphological understanding breaks down. Evaluating open-weight LLMs and frontier commercial models, we find that open-weight models perform near chance on morpheme decomposition regardless of scale. Frontier models perform much better, often recovering individual affixes under contains-match scoring, but remain far below their character-level ceilings on compositional tasks of morpheme transformations and syllabification. These results identify productive morphological composition, rather than character access alone, as the persistent bottleneck for Filipino word-structure understanding.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses limitation, benchmark.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15121unread
When Cognitive Graphs Meet LLMs: BDEI Cognitive Pathways for Panic Emotional Arousal Prediction
Mengzhu Liu, Long Qin, Chuan Ai, Zhengqiu Zhu, Hongru Liang, Chen Gao, Yong Li, Xin Lu, Quanjun Yin · 2026-06-17
arXiv:2606. 15121v2 Announce Type: new Abstract: Predicting individual panic emotional arousal timing before manifestation is essential for proactive emergency intervention.
Read next because When Cognitive Graphs Meet LLMs: BDEI Cognitive Pathways for Panic Emotional Arousal Prediction overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)". Matching terms: rect, eval, line, rate, model. Source: arxiv cs.CL (NLP).
arXiv:2606.15121v2 Announce Type: new Abstract: Predicting individual panic emotional arousal timing before manifestation is essential for proactive emergency intervention. Existing methods incorporate cognitive elements but none explicitly model the emotional arousal process, making them ill-suited for emotional arousal timing prediction. We argue that grounding prediction in appraisal emotion theory is necessary because it explicitly models this process, but three problems must be solved. (1) Appraisal theory posits that emotion arises from simultaneous evaluation across multiple threat dimensions, yet no prior work fuses these inputs into risk perception. (2) Existing cognitive models lack an Emotion node, decoupling threat appraisal from emotional arousal and forcing emotions to be inferred indirectly from behaviors. (3) Given their generalizable cognitive reasoning, current approaches adopt LLMs as the primary decision-maker, yet overlook the fragility and hallucination-proneness of their outputs. To address these issues, we introduce PanicCognitivePath (PCP), a framework that addresses all three. A Psychological Safety Distance (PSD) model, grounded in psychological distance theory, maps four-domain signals into a unified risk metric as the entry condition for subsequent cognitive reasoning. An explicit Emotion node grounded in appraisal emotion theory is introduced into BDI, forming a Belief-Desire-Emotion-Intention (BDEI) pathway. Agents whose risk metric exceeds the PSD threshold enter this pathway, coupling threat appraisal directly to emotional arousal. The BDEI pathway governs all state transitions while the LLM is confined to parameter estimation for the Belief-to-Desire transition, confining hallucinations to a single step and preventing error propagation. Experiments on Hurricane Sandy show PCP improves arousal timing accuracy by 10.68% over baselines, reduces peak count error to 7.07%.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses evaluation.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15080unread
AdaMame: A Training Recipe for Adaptive Multilingual Reasoning
Dayeon Ki, Kevin Duh, Marine Carpuat · 2026-06-17
arXiv:2606. 15080v1 Announce Type: new Abstract: While Large Reasoning Models (LRMs) show strong performance in English, they often fail to reason in the language of the query, a phenomenon known as language collapse.
Read next because AdaMame: A Training Recipe for Adaptive Multilingual Reasoning overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, strong, alignment, eval, source, token, line, recipe. Source: arxiv cs.CL (NLP).
arXiv:2606.15080v1 Announce Type: new Abstract: While Large Reasoning Models (LRMs) show strong performance in English, they often fail to reason in the language of the query, a phenomenon known as language collapse. Existing RL-based fixes typically add a binary language fidelity reward to the accuracy objective, yet still incur trade-off in accuracy, mid-trace code-switching, and excessive token usage. In this work, we propose AdaMame, a two-stage training recipe for multilingual mathematical reasoning that addresses these limitations by adaptively aligning the reasoning language to the query language without compromising accuracy. The first SFT stage fine-tunes on naturally occurring reasoning traces across five languages to establish multilingual reasoning capability. In the subsequent RL stage, we introduce AdaMame-GRPO, an adaptation of Group Relative Policy Optimization (GRPO) in which a query-conditioned alignment factor grows progressively during training, guiding the model to first explore diverse reasoning languages before exploiting reasoning in the query language. Evaluated across two benchmarks, two LRMs, and 12 languages, AdaMame-GRPO achieves Pareto-optimal performance across reasoning accuracy, language fidelity, and token efficiency over all baselines, with the strongest gains on out-of-domain, lower-resource languages.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses limitation, limitations, benchmark.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15070unread
Stop When Further Reasoning Won't Help: Attention-State Adaptive Generation in Reasoning Models
Jiakai Li, Ke Qin, Rongzheng Wang, Yizhuo Ma, Qizhi Chen, Muquan Li, Shuang Liang · 2026-06-17
arXiv:2606. 15070v1 Announce Type: new Abstract: By incorporating test-time compute scaling, large reasoning models (LRMs) can solve complex problems through explicit chain-of-thought (CoT) reasoning processes.
Read next because Stop When Further Reasoning Won't Help: Attention-State Adaptive Generation in Reasoning Models overlaps with clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)". Matching terms: source, token, rate, chain, test, model. Source: arxiv cs.CL (NLP).
arXiv:2606.15070v1 Announce Type: new Abstract: By incorporating test-time compute scaling, large reasoning models (LRMs) can solve complex problems through explicit chain-of-thought (CoT) reasoning processes. However, they often suffer from overthinking, resulting in redundant token outputs and degraded accuracy. Current methods to mitigate this issue remain limited: training-based approaches require substantial computational resources, while training-free methods rely on well-crafted prompts or unreliable confidence signals. In this work, we investigate early stopping from the perspective of attention distributions and propose a simple method, ASAG, which infers the model's reasoning state and adaptively adjusts the generation strategy. The proposed framework is training-free and plug-and-play, enabling seamless integration into existing LRMs. Extensive experiments on nine benchmarks demonstrate consistent improvements across mainstream LRMs with varying parameter scales, including the DeepSeek-R1-Distill and Qwen3 series. Specifically, ASAG improves average accuracy by 3.2% while reducing the number of generated tokens by nearly 40% across all reasoning tasks on Qwen3-8B.
Potential threat/caveat for clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)": this item discusses benchmark.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15069unread
CoCoGEC: Counterfactual Generation for Robust Grammatical Error Correction
Qianyu Wang, Xiaoman Wang, Yuanyuan Liang, Xinyuan Li, Yunshi Lan · 2026-06-17
arXiv:2606. 15069v1 Announce Type: new Abstract: Grammatical error correction (GEC) systems are usually trained and evaluated on GEC benchmarks, but their performance often drops sharply once the surrounding context is slightly perturbed or extended.
Read next because CoCoGEC: Counterfactual Generation for Robust Grammatical Error Correction overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, text, word, rect, under, correct, eval, line. Source: arxiv cs.CL (NLP).
arXiv:2606.15069v1 Announce Type: new Abstract: Grammatical error correction (GEC) systems are usually trained and evaluated on GEC benchmarks, but their performance often drops sharply once the surrounding context is slightly perturbed or extended. This indicates that the existing GEC models usually fail to understand the error patterns in the varying contexts. In this paper, we thoroughly investigate the counterfactuals for GEC tasks, where the subtle changes to the contexts could lead to the label flipping issue. We propose CoCoGEC, a counterfactual generation framework that creates copies of training instances with error-irrelevant contexts altered. Our framework systematically generates counterfactuals by (1) generating intra- and inter-sentence counterfactuals that maintain the error patterns as well as syntax of the original instances by altering the word-level and sentence-level contexts; (2) revising the generated counterfactuals by selecting the instances with flipped labels and high GEC Mutual Information (MI) coefficient. Extensive experiments show that our method substantially improves the stability of GEC models, outperforming a set of data augmentation baselines. Particularly, it could achieve absolute F0.5 gains of +9.9, +11.3, and +20.8 points on the perturbed BEA-19*,CoNLL-14*, and TEM-8* data set.Our code is released at https://github.com/Quinnok/CoCoGEC
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses benchmark.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15059unread
A Practical Evaluation Method for Long-Form Simultaneous Speech-to-Speech Translation
Yulin Xue, Siqi Ouyang, Lei Li · 2026-06-17
arXiv:2606. 15059v1 Announce Type: new Abstract: Simultaneous speech-to-speech translation (SimulS2ST) enables real-time cross-lingual communication, but existing evaluation has focused largely on short or pre-segmented speech rather than long-form, continuous input.
Read next because A Practical Evaluation Method for Long-Form Simultaneous Speech-to-Speech Translation overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, alignment, eval, source, token, rate. Source: arxiv cs.CL (NLP).
arXiv:2606.15059v1 Announce Type: new Abstract: Simultaneous speech-to-speech translation (SimulS2ST) enables real-time cross-lingual communication, but existing evaluation has focused largely on short or pre-segmented speech rather than long-form, continuous input. Prior approaches are difficult to reproduce and make assumptions that do not hold for end-to-end systems. We present a practical evaluation method for long-form SimulS2ST. Given source speech, pre-segmented source transcripts, and reference translations, we run automatic speech recognition (ASR) and forced alignment on the generated target speech to recover token-level timestamps, then apply a sentence-embedding-based aligner to match the target text to its corresponding source sentences. This enables sentence-level computation of latency and quality metrics, including YAAL and xCOMET, which are then aggregated into final system-level scores. Experiments on representative SimulS2ST systems show that the method is effective in practice and reveal that current systems suffer from substantial latency accumulation on long speech.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15044unread
Equity with Efficiency: An Empirical Study of Tokenizers for Multilingual Large Language Models
Kieron Seven Jun Wei Lee, Muhammad Reza Qorib, Andrew Ivan Soegeng, Hwee Tou Ng · 2026-06-17
arXiv:2606. 15044v1 Announce Type: new Abstract: Multilingual large language models (LLMs) depend on subword tokenization to bridge discrete text and continuous neural representation.
Read next because Equity with Efficiency: An Empirical Study of Tokenizers for Multilingual Large Language Models overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, text, word, latin, under, source, token, rate. Source: arxiv cs.CL (NLP).
arXiv:2606.15044v1 Announce Type: new Abstract: Multilingual large language models (LLMs) depend on subword tokenization to bridge discrete text and continuous neural representation. State-of-the-art multilingual LLMs often use Byte-level Byte-Pair Encoding (BPE) tokenizers that structurally favor high-resource languages and Latin scripts. For speakers of underrepresented languages, particularly those across Southeast Asia, this bias inflates inference costs and widens cross-lingual capability gaps. We present the first systematic comparison of equitable tokenizers on a unified benchmark spanning 11 Southeast Asian languages. Beyond tokenizer-level analysis of compression efficiency and cross-lingual equity, we assess downstream task performance through controlled 1.5B-parameter language model training using the same training data. Our results show that Parity-aware BPE lies on the Pareto frontier of the efficiency-equity trade-off, achieving strong compression parity at competitive cost. Morphology-Driven Byte Encoding delivers the best semantic reasoning performance through morphologically richer representations, albeit at a higher computational expense. Byte Latent Transformer underperforms on downstream tasks, possibly because its architectural assumptions misalign with the constraints of limited low-resource training data. Together, our findings demonstrate that cross-lingual fairness and tokenization efficiency are not fundamentally at odds, and offer practical guidance for designing equitable multilingual models.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses bias, benchmark.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15037unread
ReportQA: QA-Based Radiology Report Evaluation
Yiming Shi, Shaoshuai Yang, Xi Chen, Haolin Li, Hengyu Zhang, Che Jiang, Kaiwen Wang, Xun Zhu, Dong Xie, Fei Wang, Dejing Dou, Miao Li, Ji Wu · 2026-06-17
arXiv:2606. 15037v1 Announce Type: new Abstract: Radiology report evaluation is essential for advancing automated report generation.
Read next because ReportQA: QA-Based Radiology Report Evaluation overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, strong, text, rect, alignment, eval, line, rate. Source: arxiv cs.CL (NLP).
arXiv:2606.15037v1 Announce Type: new Abstract: Radiology report evaluation is essential for advancing automated report generation. Natural language generation metrics have limited clinical relevance. Clinical efficacy (CE) metrics evaluate important medical findings, but focus mainly on presence and cover only a limited set of entities. Due to heavy reliance on manual annotations, it is difficult for CE metrics to extend clinical entities or attributes. In clinical practice, radiology reports serve as a medium for information transfer. Clinicians use them to perform downstream diagnostic tasks without directly inspecting images. Based on this insight, we propose ReportQA, a clinical-related and flexible radiology report evaluation framework, supporting detailed quantitative analysis of radiology report generation systems. We first collect datasets covering multiple imaging modalities and anatomical regions. We then construct knowledge trees of clinical entities and attributes with radiologist guidance, and use large language models (LLMs) to extract structured information from raw reports. Next, we generate QA pairs from predefined templates and apply quality control through self-filtering and report-based filtering. During evaluation, the report is treated as context, and an LLM acts as a judge model to answer the QA pairs. Based on the resulting QA accuracy, we introduce QAScore metric. Compared with existing metrics, QAScore shows better alignment with radiologist judgments. Experiments on multiple state-of-the-art vision-language models reveal that current report-based inference paradigms struggle to learn fine-grained clinical representations and exhibit strong negative prior biases. In contrast, question-driven inference provides a more effective alternative. For reproducibility and extensibility, we release the knowledge trees, structured reports, and QA pairs, along with the pipeline code for QA construction and evaluation.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses bias, negative, evaluation.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15026unread
Deep Temporal Modeling and Ensemble Fusion for Multimodal Emotion Recognition from Physiological Signals
Desta Haileselassie Hagos, Saurav Keshari Aryal, Patrick Ymele-Leki, Anietie Andy, Legand L. Burge · 2026-06-17
arXiv:2606. 15026v1 Announce Type: new Abstract: Physiological stress and emotion recognition are important for health monitoring and affective computing.
Read next because Deep Temporal Modeling and Ensemble Fusion for Multimodal Emotion Recognition from Physiological Signals overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "Implement Chen et al. persona-vector extraction recipe and compare to project's centroid-difference recipe". Matching terms: eval, rate, implement, trained, model. Source: arxiv cs.CL (NLP).
arXiv:2606.15026v1 Announce Type: new Abstract: Physiological stress and emotion recognition are important for health monitoring and affective computing. In this work, we present a comprehensive evaluation of deep learning models such as Long Short-Term Memory (LSTM), Temporal Convolutional Networks (TCN), and Transformer on the WESAD dataset for multimodal affect recognition using wrist and chest sensor signals. We perform ablation studies to assess the individual contributions of each modality by training models on wrist-only and chest-only inputs. In addition, we implement a late-fusion ensemble strategy that combines predictions from all three architectures trained on multimodal input. We also employ early fusion at the sensor level by concatenating wrist and chest signals before feeding them into each model. Our results show that Transformer models consistently achieve the highest accuracy in multimodal settings, while TCN models perform best in the wrist-only configuration. The ensemble method yields the highest overall accuracy (98.91 +/- 0.13%) and macro-F1 score (98.56 +/- 0.17%). These findings demonstrate the effectiveness of sensor fusion and ensemble-based fusion in developing robust systems for physiological emotion recognition.
Potential threat/caveat for clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)": this item discusses evaluation.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.15017unread
Are Online Skill and Memory Modules Always Worth Their Tokens? A Budget-Constrained Study of Web Agents
Sina Hajimiri, Masih Aminbeidokhti, Jose Dolz, Ismail Ben Ayed, Issam H. Laradji, Spandana Gella, Nicolas Gontier · 2026-06-17
arXiv:2606. 15017v1 Announce Type: new Abstract: Online web agents often augment a base actor with memory, workflow, or skill modules.
Read next because Are Online Skill and Memory Modules Always Worth Their Tokens? A Budget-Constrained Study of Web Agents overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: under, eval, token, line, rate, compare, trained, test. Source: arxiv cs.CL (NLP).
arXiv:2606.15017v1 Announce Type: new Abstract: Online web agents often augment a base actor with memory, workflow, or skill modules. These modules can improve performance, but they also consume test-time tokens, a cost rarely reported alongside the actor's inference cost. We study online augmentation, where this overhead is paid on every task, and re-evaluate its benefits under a fixed total inference budget. We compare AWM, ASI, and ReasoningBank with a token-matched vanilla baseline that uses the same budget for additional actor steps. Across three WebArena domains and three models, Gemini 3 Flash, GPT-5.4-mini, and Qwen 3.6-27B, the vanilla baseline matches or surpasses all three augmentation methods in aggregate success rate while often using fewer total tokens. We observe a similar trend on WorkArena-L1 with Qwen 3.6-27B, indicating that the effect extends to enterprise knowledge-work tasks. Our results suggest that skills and workflow memory can be useful in specific domains, but their apparent gains often vanish against a budget-matched actor. We further show that run-to-run variance materially affects outcomes and should be reported as a core evaluation criterion for online web agents.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses evaluation.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.14867unread
Evaluating the Robustness of Proof Autoformalization in Lean 4
Zhengtao Gui, Sheng Yang, Zhouxing Shi · 2026-06-17
arXiv:2606. 14867v1 Announce Type: new Abstract: Proof autoformalization aims to translate a mathematical informal proof written in natural language into a formal proof in a formal language such as Lean~4.
Read next because Evaluating the Robustness of Proof Autoformalization in Lean 4 overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, phrase, latin, phrases, rect, under, correct, eval. Source: arxiv cs.CL (NLP).
arXiv:2606.14867v1 Announce Type: new Abstract: Proof autoformalization aims to translate a mathematical informal proof written in natural language into a formal proof in a formal language such as Lean~4. Several works have developed LLM-based models for proof autoformalization. However, existing evaluations have typically focused on translating well-formed informal proofs from curated datasets. We argue that a robust proof autoformalizer must remain faithful even for informal proofs that diverge from these idealized ones, and we present the first study on the robustness of proof autoformalization models. We formulate two categories of perturbations and evaluate robustness under each: a global perturbation paraphrases the informal proof in a different style, under which the formalization should remain consistent; a local perturbation alters a value, symbol, or proof step, possibly in a counterfactual way, and a robust formalization should faithfully reflect the perturbation rather than reverting to the original one or inferring a different one on its own. We build a benchmark with both perturbations on miniF2F and MATH-500, and automatically measure how stable a proof autoformalization's correctness is under global perturbations and how faithfully its output reflects local perturbations. We evaluate seven recent models, all of which are sensitive to global perturbations and mostly fail to remain faithful under local perturbations. Code and data are available via https://github.com/ucr-rai/robust-proof-autoformalization.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses robustness, evaluation, benchmark.
source ↗open
score 100arxiv cs.CL (NLP)arxiv:2606.14832unread
PhoneHarness: Harnessing Phone-Use Agents through Mixed GUI, CLI, and Tool Actions
Chenxin Li, Zhengyao Fang, Zhengyang Tang, Pengyuan Lyu, Xingran Zhou, Xin Lai, Fei Tang, Liang Wu, Yiduo Guo, Weinong Wang, Junyi Li, Yi Zhang, Yang Ding, Huawen Shen, Sunqi Fan, Shangpin Peng, Zheng Ruan, Anran Zhang, Benyou Wang, Chengquan Zhang, Han Hu · 2026-06-17
arXiv:2606. 14832v1 Announce Type: new Abstract: Phone agents are increasingly expected to complete real mobile workflows rather than merely predict the next screen action.
Read next because PhoneHarness: Harnessing Phone-Use Agents through Mixed GUI, CLI, and Tool Actions overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, eval, rate, control, emit, screen. Source: arxiv cs.CL (NLP).
arXiv:2606.14832v1 Announce Type: new Abstract: Phone agents are increasingly expected to complete real mobile workflows rather than merely predict the next screen action. However, much of the current mobile-agent literature still evaluates agents primarily as GUI controllers that observe a screen, emit taps and swipes, and are scored by target app state. Real phone-use tasks are broader: they require deciding when to use app GUIs, device-side commands, or structured tools, while leaving evidence that the intended side effect actually occurred. We introduce PhoneHarness, a mixed-action benchmark and execution harness for studying phone-use agents on verifiable mobile workflows. PhoneHarness runs a device-side agent loop over GUI, CLI, and host-side tool actions, combining deterministic action routing with bounded GUI delegation and auditable execution traces. Its benchmark, PhoneHarness Bench, evaluates whether agents complete tasks with observable side effects, not only whether they produce plausible final answers. On the annotated evaluation split, PhoneHarness reaches a 75.0% pass rate, outperforming the strongest non-PhoneHarness settings by 12.9 percentage points. PhoneHarness and PhoneHarness Bench therefore play distinct but mutually dependent roles: the harness makes mixed phone workflows executable, while the benchmark measures whether agents can use that harness reliably and safely. Our findings suggest that reliable phone automation depends on action-surface routing and verifiable execution, not only visual GUI control.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation, benchmark.
source ↗open
score 100arxiv cs.LG (Machine Learning)arxiv:2606.17435unread
MorphStrata: Layer-Specific Perturbations for Generating Morphence Students in Time-Series Moving Target Defense
Abhishek Bhardwaj, Arnav Doshi, Anusri Nagarajan, Thanh Quynh Nhu Ta, Mohammad Masum, Robert Chun, Jaydip Sen, Saptarshi Sengupta · 2026-06-17
arXiv:2606. 17435v1 Announce Type: new Abstract: Time-series forecasting models remain vulnerable to gradient-based adversarial attacks while existing defense mechanisms typically incur a trade-off in robustness for bounded response and compute cost.
Read next because MorphStrata: Layer-Specific Perturbations for Generating Morphence Students in Time-Series Moving Target Defense overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)". Matching terms: under, eval, line, rate, compare, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17435v1 Announce Type: new Abstract: Time-series forecasting models remain vulnerable to gradient-based adversarial attacks while existing defense mechanisms typically incur a trade-off in robustness for bounded response and compute cost. The problem is pronounced in Moving Target Defense where maintaining multiple randomized model instances substantially exacerbates the training overhead. In this work, we introduce MorphStrata, a student generation strategy with selective, layer-specific stochastic noise injection that extends the traditional Morphence defense. MorphStrata uses a Transformer backbone as the teacher and perturbs randomly selected architectural blocks to create structured heterogeneity across student models in response to varied data distributions and threat models. We evaluate against vanilla Transformer and Morphence backbones on a suite of benchmarks including the Jena Climate, Electricity Load Diagrams, and Appliances Energy Prediction using FGSM, BIM and PGD attacks across multiple attack strengths. Across datasets and attack regimes, the proposed ensemble maintains comparable adversarial RMSE. Specifically, for high entropy, periodic datasets as in the case of the AEP data, MorphStrata achieves the lowest RMSE across all attacks and perturbation budgets, improving over the static baseline by up to 24.11% and 97.97% under FGSM and BIM respectively at an epsilon value of 0.5 over 30 randomized trials. Targeting the layers to generate MorphStrata students accounts for less than 1% increase in train-times over the Morphence MTD baseline for most of the experiments, while accounting for double digit gains in adversarial RMSE reduction. We also observe a positive correlation between higher pairwise L2 distance (among generated students) and overall defense effectiveness. In summary, MorphStrata maintains adversarial robustness as an MTD defense at marginal cost deltas when compared to existing baselines.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses robustness, adversarial, benchmark.
source ↗open
score 100arxiv cs.LG (Machine Learning)arxiv:2606.17414unread
Memory-Efficient Meta-Reinforcement Learning for Adaptive Safety-Critical Control in Adversarial Spacecraft Proximity Operations
Alejandro Posadas-Nava, Richard Linares, Minduli Wijayatunga · 2026-06-17
arXiv:2606. 17414v1 Announce Type: new Abstract: Autonomous spacecraft rendezvous and proximity operations (RPO) require controllers that guarantee safety under thrust constraints while minimizing fuel expenditure.
Read next because Memory-Efficient Meta-Reinforcement Learning for Adaptive Safety-Critical Control in Adversarial Spacecraft Proximity Operations overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: class, under, soft, eval, line, compare, control, trained. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17414v1 Announce Type: new Abstract: Autonomous spacecraft rendezvous and proximity operations (RPO) require controllers that guarantee safety under thrust constraints while minimizing fuel expenditure. Input-constrained control barrier functions (ICCBFs) provide a control method for nonlinear systems with actuation constraints that construct a forward-invariant safe set. Previous work has shown that learning class-$\mathcal{K}$ functions defining the ICCBF recursion via meta reinforcement learning (meta-RL) yields a robust, non-greedy approach to safety-critical control in RPO. This paper extends that framework further by investigating the performance of three recurrent network architectures (Long Short Term Memory (LSTM), Gated Recurrent Unit (GRU), Selective State Space Model (Mamba)) and two training algorithms (Proximal Policy Optimization (PPO) and Soft Actor Critic (SAC)) to identify the best setup for tuning ICCBF class-K functions via meta-RL. In addition to cooperative test cases, performance is evaluated in the presence of adversarial behavior where the target spacecraft behaves in a way that worsens the safety of the chaser spacecraft. Results indicate that state space models such as Mamba when used with PPO achieve superior task completion, safety, and fuel-savings compared to other architectures, across all cooperative and uncooperative scenarios tested.
Potential threat/caveat for clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)": this item discusses adversarial.
source ↗open
score 100arxiv cs.LG (Machine Learning)arxiv:2606.17413unread
Amortized Probabilistic Retrieval of Atmospheric CO2 from OCO-2 Spectra Using Deep Learning with Laplace Approximations and Normalizing Flows
Alejandro Calle-Saldarriaga, Felix Jimenez, Jack Grosskreuz, Jiazheng Wang, Jonathan Hobbs, Matthias Katzfuss · 2026-06-17
arXiv:2606. 17413v1 Announce Type: new Abstract: Space-based monitoring of atmospheric carbon dioxide (CO2) is essential for constraining the global carbon budget.
Read next because Amortized Probabilistic Retrieval of Atmospheric CO2 from OCO-2 Spectra Using Deep Learning with Laplace Approximations and Normalizing Flows overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, eval, line, rate, compare, full, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17413v1 Announce Type: new Abstract: Space-based monitoring of atmospheric carbon dioxide (CO2) is essential for constraining the global carbon budget. NASA's Orbiting Carbon Observatory-2 (OCO-2) estimates column-averaged dry-air mole fractions of CO2 (XCO2) using high-resolution spectra. However, current operational retrieval algorithms are computationally expensive and do not properly quantify uncertainties. We present a novel deep learning framework that addresses these challenges. Due to the difficulties of ground-truth data for real satellite observations, we develop and validate our approach using a high-fidelity simulation dataset. This dataset, created to support OCO-2 uncertainty quantification (UQ), incorporates realistic forward model errors. Our architecture encodes spectral bands using a multi-branch neural network and estimates posteriors of the full CO2 column or desired summaries thereof using two scalable UQ methods: Laplace approximations and normalizing flows. Our approach has five key advantages relative to operational "full-physics" solvers: (1) Amortization: Inference is orders of magnitude faster, enabling real-time processing of massive data streams; (2) Model error robustness: By training on simulations that explicitly include model discrepancies, our method accounts for systematic errors often neglected by standard inversions; (3) Point estimate accuracy: We achieve superior predictive accuracy compared to baseline methods; (4) Improved UQ: The probabilistic outputs yield better-calibrated uncertainty estimates; and (5) Non-Gaussian posteriors: When utilizing normalizing flows, our framework successfully models complex, asymmetric posterior distributions, overcoming the limitations of the Gaussian assumption. These results suggest that simulation-based deep learning is a viable path toward next-generation operational processing systems.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses limitation, limitations, robustness.
source ↗open
score 100arxiv cs.LG (Machine Learning)arxiv:2606.17409unread
Discrete Autoregressive Transformer for Generative Mechanism Synthesis
Anar Nurizada, Anurag Purwar · 2026-06-17
arXiv:2606. 17409v1 Announce Type: new Abstract: Planar path synthesis requires mechanisms whose coupler curves match a prescribed trajectory; the mapping from curve to linkage is inherently one-to-many across four-, six-, and eight-bar topologies.
Read next because Discrete Autoregressive Transformer for Generative Mechanism Synthesis overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, alignment, eval, token, line, rate, without, candidates. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17409v1 Announce Type: new Abstract: Planar path synthesis requires mechanisms whose coupler curves match a prescribed trajectory; the mapping from curve to linkage is inherently one-to-many across four-, six-, and eight-bar topologies. We address this design problem with simulation-grounded evaluation on a curated corpus of over one million mechanisms, reporting Chamfer distance and dynamic time warping after forward kinematics and geometric alignment. We formulate synthesis as conditional autoregressive sequence modeling: joint coordinates are uniformly quantized to tokens and generated by a decoder-only transformer with a variational-autoencoder (VAE) latent of the target curve and an explicit mechanism-type token. Training combines token cross-entropy with a Gaussian-smoothed bin auxiliary loss that respects ordinal structure among bins. At inference, a bounded latent-noise schedule decodes all mechanism types at each noise level; we retain the top five candidates by geometric error, yielding diverse accurate families without dataset lookup. On held-out tests, aggregate mean Chamfer distance is $0.0132$ and mean dynamic time warping is $0.153$; a latent $k$-nearest-neighbor baseline that conditions on training-set neighbor latents in VAE space achieves matched-topology mean Chamfer distance $0.0071$ and mean dynamic time warping $0.117$ using the same decoder.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation.
source ↗open
score 100arxiv cs.LG (Machine Learning)arxiv:2606.17331unread
Decision-Driven Geosteering Under Uncertainty: A Unified Framework for Sequential Decision Optimization
Hibat Errahmen Djecta, Sergey Alyaev, Kristian Fossum, Reidar B. Bratvold, Ressi Bonti Muhammad, Apoorv Srivastava · 2026-06-17
arXiv:2606. 17331v1 Announce Type: new Abstract: Geosteering requires navigating a well trajectory through an unknown geological configuration, while sequentially updating decisions based on indirect measurements acquired during drilling.
Read next because Decision-Driven Geosteering Under Uncertainty: A Unified Framework for Sequential Decision Optimization overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: rect, under, correct, eval, line, rate, control, trained. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17331v1 Announce Type: new Abstract: Geosteering requires navigating a well trajectory through an unknown geological configuration, while sequentially updating decisions based on indirect measurements acquired during drilling. This work presents an uncertainty-aware geosteering framework that tightly integrates particle filtering for probabilistic subsurface interpretation with value-based reinforcement learning for sequential decision-making. Geological uncertainty ahead of the drill bit is represented explicitly through a particle filter (PF), enabling belief-informed control rather than deterministic trajectory correction. The framework couples PF belief updates with belief-informed decision policies and evaluates three decision-making options that operate under identical uncertainty representations: an interpretable Approximate Dynamic Programming (ADP) scheme, a Deep Q-learning baseline, and a Dual Deep Reinforcement Learning (Dual DRL) architecture trained with a target Q-network scheme for stability, using a dueling (value/advantage) decomposition for Q-value parameterization. Beyond final placement performance, we assess policy behavior using stability-oriented metrics that quantify steering smoothness over time, providing additional operational insight into how decision policies respond as uncertainty evolves. The framework is integrated with an API for validation within an industrial geosteering simulator under realistic measurement noise and drilling constraints. Using identical geological realizations, operational limits, and reward definitions across methods, the experiments provide a controlled and high-fidelity evaluation of how alternative decision policies behave throughout the drilling process, rather than evaluating performance solely from the final well trajectory.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses evaluation.
source ↗open
score 100arxiv cs.LG (Machine Learning)arxiv:2606.17321unread
ProCUA-SFT Technical Report
Jaehun Jung, Ximing Lu, Brandon Cui, Muhammad Khalifa, Shaokun Zhang, Hao Zhang, Jin Xu, Amala Sanjay Deshmukh, Karan Sapra, Andrew Tao, Yejin Choi, Jan Kautz, Mingjie Liu, Yi Dong · 2026-06-17
arXiv:2606. 17321v1 Announce Type: new Abstract: Training computer-use agents (CUAs) -- models that interact with graphical desktops through screenshots and keyboard/mouse actions -- requires large-scale, diverse trajectory data collected in full desktop environments.
Read next because ProCUA-SFT Technical Report overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, prefix, source, line, rate, full, trained, screen. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17321v1 Announce Type: new Abstract: Training computer-use agents (CUAs) -- models that interact with graphical desktops through screenshots and keyboard/mouse actions -- requires large-scale, diverse trajectory data collected in full desktop environments. The largest public resource, AgentNet (22.5K human trajectories), leads to negative transfer when used for supervised fine-tuning (SFT): continuing training UI-TARS 7B on AgentNet causes OSWorld success rate to fall from 26.3% to 8-10%. We present ProCUA-SFT, a dataset of 3.1M step-level SFT samples distilled from 93K synthetic trajectories across 2,484 application combinations. The dataset is produced by a fully automated pipeline that (i) synthesizes grounded tasks on live desktops seeded with real-world content -- 912 spreadsheets from SpreadsheetBench, approximately 10K permissively-licensed presentations from Zenodo10K, and multi-application OSWorld configs -- and (ii) verifies each task's feasibility through binary precondition checking before rollout. A single VLM (Kimi-K2.5) serves as goal generator, precondition judge, and trajectory executor, eliminating planner-actor capability gaps. Each trajectory is expanded into step-prefix samples that exactly reproduce the context layout seen at inference time. Fine-tuning UI-TARS 7B on ProCUA-SFT for one epoch yields 45.0% on OSWorld -- an 18.7 percentage-point improvement over the base model and over 35% above AgentNet-trained counterparts. A subset of ProCUA was incorporated into the training data for the Nemotron 3 Nano Omni model, contributing to its computer-use capabilities.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses negative.
source ↗open
score 100arxiv cs.LG (Machine Learning)arxiv:2606.17250unread
Rethinking Groups in Critic-Free RLVR
Yihong Wu, Liheng Ma, Lingfeng Xiao, Muzhi Li, Xinyu Wang, Yingxue Zhang, Jian-Yun Nie · 2026-06-17
arXiv:2606. 17250v1 Announce Type: new Abstract: Reinforcement learning (RL) has become a central paradigm for post-training large language models.
Read next because Rethinking Groups in Critic-Free RLVR overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, under, token, line, rate, language, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17250v1 Announce Type: new Abstract: Reinforcement learning (RL) has become a central paradigm for post-training large language models. Existing critic-free RL methods typically generate a group of rollouts for the same question to estimate value baselines for advantage computation. However, this design suffers from data inefficiency, group synchronization barriers, and inflexibility with structured rollouts. In this work, we revisit the role of the ``group'' and show that its underlying function is not merely to estimate baselines but to prevent false penalties on negative samples. Building on this insight, we propose negative token filtering, a simple and effective strategy that enables stable single-rollout training. We apply it to two batch-level advantage methods, achieving comparable performance on reasoning tasks and stronger performance on agentic tasks relative to group-based RL techniques.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses negative.
source ↗open
score 100arxiv cs.LG (Machine Learning)arxiv:2606.17229unread
Rift: A Conflict Signature for Deception in Language Models
Petr Nyoma · 2026-06-17
arXiv:2606. 17229v1 Announce Type: new Abstract: A model that lies while knowing the truth is the central case ELK cannot handle with behavioral evaluation alone.
Read next because Rift: A Conflict Signature for Deception in Language Models overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, rect, correct, wrong, eval, rate, control, alone. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17229v1 Announce Type: new Abstract: A model that lies while knowing the truth is the central case ELK cannot handle with behavioral evaluation alone. We ask whether such deception leaves an internal signature distinguishing it from honest error. Our key move is a control for wrongness: we contrast a sleeper agent (knows the truth, lies on trigger) against a naive liar (fine-tuned to emit the same wrong answers with no honest training). Both produce identical wrong outputs; any difference is about knowledge conflict, not incorrectness. We find deceptive forward passes carry a conflict signature - 2.1-2.3x higher residual rank than naive-liar passes on the same wrong answer - strong enough to identify which of two responses is the lie with 100% accuracy and no labels, across GPT-2 small/medium (three seeds) and three instruct models. Across Qwen2.5-1.5B/7B and Phi-3-mini, instructed deception raises residual rank on every tested fact (18/18, 40/40, 34/34); on Phi-3, lies separate perfectly from both honest answers and hallucinations (AUC 1.0, Wilcoxon p~6e-11). The signature survives strategic self-constructed deception (model invents its own lie, AUC 1.0), active concealment attempts (AUC 1.0), and length-controlled replication (20/20, AUC 1.0, p~1e-6). Using basis-free relative representations, a probe trained on one model family detects deception in two other families zero-shot (mean AUC 0.933), surviving simultaneous architecture and format change (AUC 0.821), and transfers across five languages (AUC 1.000, length-controlled). The signature is read-only: detectable but not injectable (0/8 both directions). Honest limitations and six negative experiments are documented in full.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses limitation, limitations, negative, evaluation.
source ↗open
score 100arxiv cs.LG (Machine Learning)arxiv:2606.17199unread
PowerOPD: Stabilizing On-Policy Distillation with Bounded Power Transformation
Anhao Zhao, Junlong Tong, Yingqi Fan, Ping Nie, Wenjie Li, Xiaoyu Shen · 2026-06-17
arXiv:2606. 17199v1 Announce Type: new Abstract: Standard on-policy distillation (OPD) for large language models estimates the reverse-KL objective using student-sampled tokens, yielding an unbiased single-sample Monte Carlo estimator that avoids vocabulary-wide computation.
Read next because PowerOPD: Stabilizing On-Policy Distillation with Bounded Power Transformation overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)". Matching terms: alpha, token, rate, compare, full, on-policy, position, language. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17199v1 Announce Type: new Abstract: Standard on-policy distillation (OPD) for large language models estimates the reverse-KL objective using student-sampled tokens, yielding an unbiased single-sample Monte Carlo estimator that avoids vocabulary-wide computation. However, we show that this estimator suffers from severe training pathologies in practice: sample inefficiency, unstable generation dynamics, and a substantial performance gap compared to exact full-vocabulary OPD. Reward-level diagnosis traces these pathologies to the log-ratio reward, which is unbounded by construction, producing extremely high-variance gradients concentrated at early positions and persisting throughout training; standard post-hoc scaling fail as they operate only after this distortion occurs. To solve this problem, we propose PowerOPD: a family of natively bounded, sign-consistent rewards from the Box-Cox power transformation, parameterized by alpha > 0, of which the log-ratio is the degenerate alpha -> 0 limit. Across six mathematical reasoning benchmarks and four Qwen3 teacher-student pairs, PowerOPD achieves benchmark-averaged Avg@8/Pass@8 gains of up to +6.37/+5.71 over vanilla OPD, +3.01/+3.54 over post-hoc stabilization, and +2.59/+8.90 over full-vocabulary OPD, while reducing wall-clock time by 59.2% and peak GPU memory by 23.1%. Larger alpha generally improves accuracy, consistently shortens responses, and keeps gradient norms more than 3,000x smaller than vanilla OPD.
Potential threat/caveat for clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)": this item discusses bias, benchmark.
source ↗open
score 100arxiv cs.LG (Machine Learning)arxiv:2606.17180unread
Towards Fast GNN Surrogates for CO2 Migration in Complex Geological Formations
Rodrigo S. Luna, Thiago H. N. Coelho, Luiz S. L. Neto, Roberto M. Velho, Adriano M. A. Cortes, Renato N. Elias, Alexandre G. Evsukoff, Fernando A. Rochinha, Mauricio Araya-Polo, Herve Gross, Alvaro L. G. A. Coutinho · 2026-06-17
arXiv:2606. 17180v1 Announce Type: new Abstract: This chapter discusses how a data-driven machine learning approach can reproduce key aspects of the physical behavior of multiphase flows in complex geological formations.
Read next because Towards Fast GNN Surrogates for CO2 Migration in Complex Geological Formations overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, strong, rect, eval, rate, trained, test, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17180v1 Announce Type: new Abstract: This chapter discusses how a data-driven machine learning approach can reproduce key aspects of the physical behavior of multiphase flows in complex geological formations. We propose an end-to-end graph neural surrogate tailored to CO$_2$ plume migration forecasting in geological storage. The method is evaluated on the SPE11A benchmark, a well-known industry test case designed to assess CO$_2$ storage scenarios and characterized by sharp gas-water interfaces, strong advective transport, and rapid convective mixing with fingering development. The benchmark is reformulated as a graph in which nodes represent computational cells and edges encode transmissibility-based interactions enriched with geometric attributes. Directional transport arising from grid geometry, permeability contrasts, and geological heterogeneity is captured through an anisotropic message-passing mechanism, where interaction weights are computed via geometry-conditioned edge embeddings, biasing message aggregation toward physically relevant transport directions. Temporal evolution is modeled in latent space using an autoregressive residual formulation trained with multi-step supervision. The proposed model produces competitive forecasts of gas saturation and liquid-phase density, which are key indicators for CO$_2$ storage monitoring, with cumulative errors that remain moderate over extended forecasting horizons.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses bias, benchmark.
source ↗open
score 100arxiv cs.LG (Machine Learning)arxiv:2606.17118unread
MODE: Modality-Decomposed Expert-Level Mixed-Precision Quantization for MoE Multimodal LLMs
Yuanteng Chen, Peisong Wang, Zhilei Liu, Nanxin Zeng, Yuantian Shao, Shiqiang Lang, Tao Liu, Chuangyi Li, Qinghao Hu, Gang Li, Jing Liu, Jian Cheng · 2026-06-17
arXiv:2606. 17118v1 Announce Type: new Abstract: Mixture-of-Experts Multimodal Large Language Models (MoE-MLLMs) offer remarkable performance but incur prohibitive GPU memory costs, making compression essential.
Read next because MODE: Modality-Decomposed Expert-Level Mixed-Precision Quantization for MoE Multimodal LLMs overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, under, width, eval, token, line, rate, language. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17118v1 Announce Type: new Abstract: Mixture-of-Experts Multimodal Large Language Models (MoE-MLLMs) offer remarkable performance but incur prohibitive GPU memory costs, making compression essential. Among PTQ methods, expert-level mixed-precision quantization has proven effective for MoE-LLMs, yet suffers notable degradation on MoE-MLLMs due to two overlooked biases in expert importance estimation. (1) At the cross-modal level, the numerical dominance of vision tokens causes expert selection frequency to be dominated by vision tokens, masking experts that are critical to the text modality; (2) at the intra-vision level, the large proportion of redundant vision tokens further skew frequency statistics, obscuring experts critical for informative visual content. To bridge gaps, we propose MODE, a modality-decomposed expert-level mixed-precision quantization framework for MoE-MLLMs that decomposes expert selection frequency by modality, filters redundant vision tokens to obtain denoised visual frequency, and further evaluates quantization sensitivity per modality as a complementary signal to frequency-based estimation. These signals are integrated into an Integer Linear Programming formulation to assign per-expert bit-widths under a given budget. Extensive experiments show that MODE is particularly well-suited for MoE-MLLMs, limiting average performance loss to within 2.9% at W3A16, with larger gains at the extreme 2-bit setting.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses bias.
source ↗open
score 100arxiv cs.LG (Machine Learning)arxiv:2606.17115unread
Probing, Fusion, and Trustworthiness: A Systematic Evaluation of Foundation Model Representations for Multimodal Cancer Analysis
Jingyu Hu, Giuseppe Tripodi, Reed Naidoo, Sarah F. McGough, Tapabrata Chakraborti · 2026-06-17
arXiv:2606. 17115v1 Announce Type: new Abstract: Foundation models (FMs) have emerged as powerful representation extractors for medical data, yet their generalizability to datasets under distribution shift remains underexplored.
Read next because Probing, Fusion, and Trustworthiness: A Systematic Evaluation of Foundation Model Representations for Multimodal Cancer Analysis overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: class, under, eval, line, rate, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17115v1 Announce Type: new Abstract: Foundation models (FMs) have emerged as powerful representation extractors for medical data, yet their generalizability to datasets under distribution shift remains underexplored. This work systematically evaluates FM-based representations on a suite of computational pathology tasks across two real-world commercial cohorts, IH-BC and IH-NSCLC, drawn from the licensed in-house (IH) oncology dataset. The analysis focuses on two modalities, whole-slide images and transcriptomic profiles, drawn from the IH multimodal data. We first benchmark unimodal probing performance across five FMs on eight downstream classification tasks, and find that image and omics representations carry complementary predictive signals. Then we investigate whether multimodal fusion can yield additional gains over unimodal baselines by comparing three image-omics fusion strategies built on paired representations. The trustworthiness of selected unimodal and multimodal pipelines is further assessed through conformal prediction. Our results show that FM representations achieve competitive performance on out-of-distribution data and that multimodal fusion helps mainly when no single modality dominates the signal. Conformal prediction reveals that in the majority of cases where a point prediction fails, the true diagnosis remains recoverable within the prediction set, reinforcing the value of uncertainty-aware inference for clinical support.
Potential threat/caveat for clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)": this item discusses evaluation, benchmark.
source ↗open
score 100arxiv cs.LG (Machine Learning)arxiv:2606.17113unread
The Critical Role of Model Selection in Causal Inference: A Comparative Analysis of Classification Models within the InferBERT Framework for Pharmacovigilance
Csaba Kiss, Roland Molontay, Gabriele Pergola · 2026-06-17
arXiv:2606. 17113v1 Announce Type: new Abstract: Distinguishing causal adverse drug events (ADEs) from spurious correlations remains a central challenge in pharmacovigilance.
Read next because The Critical Role of Model Selection in Causal Inference: A Comparative Analysis of Classification Models within the InferBERT Framework for Pharmacovigilance overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, class, under, eval, line, rate, test, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17113v1 Announce Type: new Abstract: Distinguishing causal adverse drug events (ADEs) from spurious correlations remains a central challenge in pharmacovigilance. The InferBERT framework integrates transformer models with Do-calculus, but its success hinges on the underlying classification model. This study evaluates the impact of model choice in InferBERT, assessing whether simpler models suffice, if domain-specific pre-training helps, whether scaling to LLMs improves causal detection, and the effect of post-hoc calibration. We performed a comparative study on two benchmarks: Analgesics-induced Acute Liver Failure (AILF) and Tramadol-related Mortalities (TRAM). Four models were evaluated-XGBoost (baseline), ALBERT (original InferBERT), BioBERT (biomedical transformer), and Med-LLaMA (medical LLM)-using 5-fold cross-validation repeated over 20 runs. We measured accuracy, Expected Calibration Error (ECE) pre- and post-isotonic regression, and Jaccard concordance of causal terms with PRR, ROR, and EBGM; significance was tested with paired t-tests. BioBERT achieved the highest accuracy on both datasets, while Med-LLaMA underperformed despite its size and parameter-efficient fine-tuning. Domain-specific pre-training was decisive. Calibration improved ECE but had mixed effects on accuracy and causal discovery. BioBERT's superiority also yielded the strongest concordance with traditional pharmacovigilance signals. These results show that domain-specific pre-training provides a clear advantage over simpler baselines and larger LLMs. Investing in manageable, domain-aware models is more effective for computational pharmacovigilance than simply scaling model size.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses failure, benchmark.
source ↗open
score 100arxiv cs.LG (Machine Learning)arxiv:2606.17107unread
Models Take Notes at Prefill: KV Cache Can Be Editable and Composable
Bojie Li · 2026-06-17
arXiv:2606. 17107v1 Announce Type: new Abstract: Prefix caching reuses prefill only across an exactly shared prefix, so one changed field invalidates the entire downstream cache.
Read next because Models Take Notes at Prefill: KV Cache Can Be Editable and Composable overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, fill, under, prefix, token, line, rate, without. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17107v1 Announce Type: new Abstract: Prefix caching reuses prefill only across an exactly shared prefix, so one changed field invalidates the entire downstream cache. Yet overwriting the field's own key/value vectors and reusing the rest leaves the model acting on the old value. The reason, established causally across four model families: at prefill the model has already written the field-conditioned conclusion onto downstream notes; the field's own key/value drives under 1% of the decision. Read as a notebook of memoized conclusions, two capabilities follow. (1) It is editable. A salient erratum amends the notes; and with chain-of-thought, editing the field alone recovers the decision (1.00 at 8B, ~1% compute), while without CoT it is ignored. (2) It is composable. The notes are position-portable, so a precompiled skill can be RoPE-repositioned and spliced into any context, indistinguishable from full recompute (logit cosine 0.90-0.999, twelve models) at O(L) rather than O(L^2) time-to-first-token. A unified edit+compose agent stays decision-identical to recompute at up to 14.9x lower latency. The approach applies to any per-token attention KV cache, validated across scale, quantization, Mixture-of-Experts, and multimodal caches, and extends to several attention variants through small adapters. Because the erratum is append-only, it composes with production prefix caching: in an online vLLM benchmark it keeps the prefix cache-aligned (98.5% hit-rate), cutting p90 time-to-first-token by 53-398x.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses benchmark.
source ↗open
score 100arxiv cs.LG (Machine Learning)arxiv:2606.17106unread
Informative Missingness to Generate Irregular Clinical Time Series
Hadi Mehdizavareh, Gabriele Santangelo, Giovanna Nicora, Simon Lebech Cichosz, Arianna Dagliati, Arijit Khan, Riccardo Bellazzi · 2026-06-17
arXiv:2606. 17106v1 Announce Type: new Abstract: Laboratory tests in electronic health records are collected irregularly, and the absence of a test order can be as informative as the measurement itself.
Read next because Informative Missingness to Generate Irregular Clinical Time Series overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: rect, under, rate, test, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17106v1 Announce Type: new Abstract: Laboratory tests in electronic health records are collected irregularly, and the absence of a test order can be as informative as the measurement itself. Such missingness reflects clinicians' decisions and patient physiology, making it important to model it directly rather than treat it as a preprocessing artifact. Here we present a diffusion-based approach for generating clinical time series that jointly models laboratory values and their observation patterns using the public Data Analytics Challenge on Missing Data Imputation (DACMI) benchmark derived from MIMIC-III. To preserve realistic sampling, we align chart times into 4-hour intervals and segment admissions into 7-day windows, producing trajectories that pair each lab value with a corresponding observation indicator. Standard transformations and normalization are applied to stabilize training. Our method extends the TimeDiff framework to learn continuous lab values and discrete missingness patterns through complementary diffusion objectives. Experiments show that the generated data closely match real patient trajectories across individual lab distributions and joint value-missingness embeddings, demonstrating that diffusion models can capture clinically meaningful dependencies between patient physiology and clinicians' testing behavior under MNAR-like (missing-not-at-random) missingness. These preliminary results indicate that our model can serve as an initial component toward developing clinical foundation models. By producing synthetic priors that preserve key physiology-missingness relationships, this work motivates the subsequent training of Prior-Data Fitted Networks capable of leveraging informative missingness, which we will investigate in the extended work.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses benchmark.
source ↗open
score 100arxiv cs.LG (Machine Learning)arxiv:2606.17093unread
Diagnosing and Repairing Shape-Prior Shortcuts in Long-Range Single-Shot Fringe Projection Profilometry
Adam Haroon, Anush Lakshman, Cody Fleming, Beiwen Li · 2026-06-17
arXiv:2606. 17093v1 Announce Type: new Abstract: Learning-based single-shot fringe projection profilometry (FPP) has been studied mostly at close range.
Read next because Diagnosing and Repairing Shape-Prior Shortcuts in Long-Range Single-Shot Fringe Projection Profilometry overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)". Matching terms: latin, soft, line, project, factor, test, absent. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17093v1 Announce Type: new Abstract: Learning-based single-shot fringe projection profilometry (FPP) has been studied mostly at close range. The long-range regime (standoff beyond 1 m) remains largely unaddressed: inverse-square intensity falloff lowers fringe signal-to-noise ratio and degrades physical ground truth, the single-shot problem is ill-posed because fringe-order information is absent from one image, and these architectures have not been studied mechanistically. We present a diagnose-repair-verify study using mechanistic interpretability (MI) and conformal uncertainty quantification (UQ) as convergent diagnostics: they agree on one physical failure locus, driving and verifying an architectural repair. On a photorealistic synthetic benchmark (15,600 fringe images, 50 objects at 1.5-2.1 m), a best UNet baseline reaches 14.54 mm object mean absolute error (MAE). Three probes (linear probing, Grad-CAM, flat-plane out-of-distribution test) converge: the baseline solves the task via object-boundary shape priors rather than fringe-phase decoding. We repair this with PhiCalNet, which outputs wrapped phase rather than depth and applies a fixed differentiable calibration layer mapping phase to depth, removing the shape-prior solution from the hypothesis space architecturally rather than by a loss penalty. A physics-informed loss that enforces the same physics as a soft penalty on a depth-regressing network yields no measurable gain, isolating the architecture as the operative factor. PhiCalNet reduces object MAE 3.3x to 4.46 mm; the residual is carried by 0.103% of pixels at the +/-pi wrap discontinuity. Pixel-wise conformal UQ confirms the diagnosis: rejecting the top 5% of object pixels by snapshot disagreement cuts PhiCalNet RMSE by 64% (20.6->7.4 mm) versus 3.5% for the baseline. MI and UQ converge on the same failure locus.
Potential threat/caveat for clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)": this item discusses failure, benchmark.
source ↗open
score 100arxiv cs.LG (Machine Learning)arxiv:2606.17057unread
Correct When Paired, Wrong When Split: Decoupling and Editing Modality-Specific Neurons in MLLMs
Tingchao Fu, Wenkai Wang, Fanxiao Li, Huadong Zhang, Jinhong Zhang, Dayang Li, Yunyun Dong, Renyang Liu, Wei Zhou · 2026-06-17
arXiv:2606. 17057v1 Announce Type: new Abstract: Although Knowledge Editing provides an efficient mechanism for updating the knowledge of Multimodal Large Language Models (MLLMs), we find that current paradigms still suffer from an important yet remain underexplored issue : editing decoupling failure, where entity-related knowledge can be updated when the model is triggered by multimodal inputs (text--image query pairs), however, it often reverts to outdated pre-edit facts when the paired inputs are split into unimodal ones.
Read next because Correct When Paired, Wrong When Split: Decoupling and Editing Modality-Specific Neurons in MLLMs overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, text, rect, under, correct, wrong, rate, propagate. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17057v1 Announce Type: new Abstract: Although Knowledge Editing provides an efficient mechanism for updating the knowledge of Multimodal Large Language Models (MLLMs), we find that current paradigms still suffer from an important yet remain underexplored issue : editing decoupling failure, where entity-related knowledge can be updated when the model is triggered by multimodal inputs (text--image query pairs), however, it often reverts to outdated pre-edit facts when the paired inputs are split into unimodal ones. Our in-depth empirical analysis reveals that the entity knowledge in MLLMs is not stored as a unified representation, but is instead distributed across disentangled modality-specific pathways. As a result, updates biased toward multimodal queries fail to propagate effectively to unimodal circuits. To bridge this gap, we propose DECODE, which explicitly disentangles and localizes modality-specific neuron groups for targeted knowledge. Extensive experiments demonstrate that DECODE consistently achieves effective knowledge updates under different modality triggers, thereby mitigating editing decoupling failures.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses failure, failures, bias.
source ↗open
score 100arxiv stat.ML (Machine Learning)arxiv:2606.17916unread
Nested Sampling: A Critical and Comprehensive Theoretical Guide
Luca Martino, Fernando Llorente · 2026-06-17
arXiv:2606. 17916v1 Announce Type: cross Abstract: The nested sampling (NS) technique has gained widespread attention, particularly in cosmology and astronomy, due to its ability to efficiently explore high-likelihood regions - a feature akin to an implicit likelihood optimization that underlies its success.
Read next because Nested Sampling: A Critical and Comprehensive Theoretical Guide overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", experiment "Implement Chen et al. persona-vector extraction recipe and compare to project's centroid-difference recipe", experiment "Follow-up to #354: cascading chunk-binding — does A→B, B→C, C→D propagate the full chain on a recipient trained only to emit A?". Matching terms: under, implement, full, trained, position. Source: arxiv stat.ML (Machine Learning).
arXiv:2606.17916v1 Announce Type: cross Abstract: The nested sampling (NS) technique has gained widespread attention, particularly in cosmology and astronomy, due to its ability to efficiently explore high-likelihood regions - a feature akin to an implicit likelihood optimization that underlies its success. While the full theoretical derivation of NS is complex and involves several approximations, the central challenge lies in sampling from the likelihood-constrained priors, which is crucial for its performance. This work provides a comprehensive and detailed exposition of NS derivation, clarifying both its theoretical foundations and practical challenges. We provide a thorough description of the NS procedure, emphasizing both its strengths and potential limitations. In doing so, this work seeks to deepen understanding of the method and to foster the development of future enhancements, novel variants, and more efficient implementations across a wide range of scientific applications. Thus, the main contribution of this work is twofold: it serves both as a tutorial for newcomers to the field and as a critical review for experienced practitioners.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses limitation, limitations.
source ↗open
score 100arxiv stat.ML (Machine Learning)arxiv:2606.17600unread
Proximal Mediation Analysis with Hidden Recanting Witnesses
Sihan Wu, Yang Bai, Yifan Cui · 2026-06-17
arXiv:2606. 17600v1 Announce Type: cross Abstract: Mediation analysis is essential for decomposing the causal effect of a treatment into direct and indirect pathways.
Read next because Proximal Mediation Analysis with Hidden Recanting Witnesses overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)". Matching terms: rect, under, correct, rate, full, model, absent. Source: arxiv stat.ML (Machine Learning).
arXiv:2606.17600v1 Announce Type: cross Abstract: Mediation analysis is essential for decomposing the causal effect of a treatment into direct and indirect pathways. However, many practical settings rely on the stringent assumption that recanting witnesses, defined as treatment-induced mediator-outcome confounders, are either absent or fully known a priori. Such a requirement is often untenable, especially when these variables remain unobservable due to measurement difficulties or privacy constraints. In this paper, we leverage proximal causal inference to develop three novel identification strategies to address the challenge of identifying path-specific effects in the presence of unknown recanting witnesses. Building on this, we develop a semiparametric inference framework that derives the efficient influence function and proposes a proximal multiply robust estimator, which remains consistent if at least one set of nuisance models is correctly specified. When all nuisance models are correctly specified and converge at appropriate rates, the estimator is asymptotically normal and achieves the semiparametric efficiency bound. We provide a minimax optimization-based debiased machine learning procedure for point estimation and constructing valid confidence intervals. The performance of the proposed methods is demonstrated by simulation studies and a real data application.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses bias, confound.
source ↗open
score 100arxiv stat.ML (Machine Learning)arxiv:2606.17516unread
FoundCause: Causal Discovery with Latent Confounders from Observational Data
Patrick Bl\"obaum, Krishnakumar Balasubramanian, Shiva Prasad Kasiviswanathan · 2026-06-17
arXiv:2606. 17516v1 Announce Type: cross Abstract: Causal discovery from observational data remains challenging due to the need to recover directed structure and latent confounding without interventions.
Read next because FoundCause: Causal Discovery with Latent Confounders from Observational Data overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, strong, class, rect, under, token, rate, without. Source: arxiv stat.ML (Machine Learning).
arXiv:2606.17516v1 Announce Type: cross Abstract: Causal discovery from observational data remains challenging due to the need to recover directed structure and latent confounding without interventions. We propose FoundCause, an amortized causal discovery model trained entirely on synthetic data that maps datasets directly to causal graphs in a single forward pass. By learning from large collections of simulated structural causal models, FoundCause captures transferable statistical patterns that generalize beyond individual datasets. The architecture incorporates several key inductive biases for causal discovery. It uses a permutation-invariant transformer encoder with alternating attention over samples and variables to jointly model cross-variable dependence and per-variable distributions. Pairwise statistical features derived from classical asymmetry measures are injected through statistics-conditioned attention, guiding the model toward known causal signals. A factorized decoder separates edge existence from direction, while a triangular refinement module enables reasoning over higher-order causal motifs such as chains and colliders. In addition, a dedicated confounder module based on learnable latent tokens explicitly models hidden common causes, and the model explicitly handles missing data via its masked input representation. To our knowledge, FoundCause is the first amortized causal discovery approach to explicitly model latent confounding. FoundCause outperforms 11 classical non-amortized methods (e.g., PC, GES, NOTEARS-style optimization) and 4 amortized causal discovery methods on 15 real-world datasets, achieving +9.6% improvement in $F_1$, +1.2% in AUROC, and an 18.9% reduction in structural Hamming distance relative to the strongest non-amortized methods, while performing inference in a single forward pass.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses bias, confound.
source ↗open
score 100arxiv stat.ML (Machine Learning)arxiv:2606.17423unread
Martingale Doppelg\"anger-Eval: An Identification Framework for Auditing Candlestick Understanding in Vision-Language Models
Ziyao Wang · 2026-06-17
arXiv:2606. 17423v1 Announce Type: cross Abstract: We introduce Martingale Doppelg\"anger-Eval, a public shadow-market benchmark for auditing whether vision-language models (VLMs) use candlestick evidence rather than extrapolate past trends.
Read next because Martingale Doppelg\"anger-Eval: An Identification Framework for Auditing Candlestick Understanding in Vision-Language Models overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, rect, under, alpha, eval, rate, control, test. Source: arxiv stat.ML (Machine Learning).
arXiv:2606.17423v1 Announce Type: cross Abstract: We introduce Martingale Doppelg\"anger-Eval, a public shadow-market benchmark for auditing whether vision-language models (VLMs) use candlestick evidence rather than extrapolate past trends. The central difficulty is identification: on real market histories, chart evidence and trend are strongly coupled, so an observational score cannot determine whether a fluent technical-analysis narrative is grounded in local visual evidence. We prove this limitation formally: no evaluation functional computed from observational chart--label data can distinguish a grounded responder from a trend-shortcut responder under strong coupling, whereas matched evidence interventions separate the same responders at an exponential rate and trend--label swaps provide an independent shortcut stress test. The benchmark therefore evaluates frozen VLMs on rendered OHLCV charts under four controlled mechanisms: a martingale-null market, injected-alpha counterfactual pairs, trend-confounder swaps, and regime shifts. A structural behavioral model identifies null-market bias, trend sensitivity, evidence sensitivity, prompt/renderer fragility, and evidence faithfulness; the accompanying statistical toolkit provides minimum detectable effects, block-aware sequential testing for metered APIs, and an overlap-weighted artifact check. Across frozen commercial and open VLMs, the identified regression assigns large positive coefficients to past trend but evidence coefficients that are zero or opposite to the rule-implied sign. Matched-pair analyses show that models either ignore injected candlestick semantics or move opposite to the rule-implied direction conditional on responding. The benchmark isolates a failure mode that standard observational chart benchmarks cannot detect and gives a reusable audit template for time-series imagery with controllable label mechanisms.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses failure, limitation, bias, confound, evaluation.
source ↗open
score 100arxiv stat.ML (Machine Learning)arxiv:2606.17233unread
Uncertainty Quantification of Engineering Structures by Polynomial Chaos Expansion and Multivariate Active Learning
Qitian Lu, Jafar Jafari-Asl, Panagiotis Spyridis, Lukas Novak · 2026-06-17
arXiv:2606. 17233v1 Announce Type: new Abstract: In many engineering applications, a single high-fidelity model produces multiple quantities of interest (QoIs) under the same input parameters, e.
Read next because Uncertainty Quantification of Engineering Structures by Polynomial Chaos Expansion and Multivariate Active Learning overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, latin, rect, under, eval, rate, compare, candidate. Source: arxiv cs.LG (Machine Learning).
arXiv:2606.17233v1 Announce Type: new Abstract: In many engineering applications, a single high-fidelity model produces multiple quantities of interest (QoIs) under the same input parameters, e.g. finite element models of complex physical systems. To alleviate the high computational cost of direct model evaluations, surrogate models are widely used to construct efficient approximations of model responses. Naturally, the accuracy of surrogates strongly depends on the quality of the experimental design (ED). However, a single ED may not provide an adequate representation for all outputs simultaneously, especially when different outputs exhibit varying sensitivities to the input variables. A straightforward solution is to perform separate sampling for each output, but this results in increased sampling complexity and computational cost. From a statistical perspective, such an approach also ignores potential correlations among all outputs and may compromise data consistency. To address this issue, an adaptive sequential sampling method for constructing polynomial chaos expansion surrogate models is generalized for vector valued QoIs. The method sequentially selects new samples from a candidate pool based on their local contribution to the output variance, while balancing distance-based exploration of the input space and exploitation of aggregated variance information across all outputs. Its performance is compared with non-sequential Latin Hypercube Sampling through several numerical examples from engineering problems. Numerical results demonstrate that the proposed strategy improves both surrogate accuracy and stability, and provides a more reliable estimation of second-order statistics.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation.
source ↗open
score 100arxiv stat.ML (Machine Learning)arxiv:2606.18183unread
A Diffusion Approximation for Temporal-Difference Learning with Linear Features under Markovian Noise
M. Forzo, E. Monzio Compagnoni, A. Russo, A. Pacchiano · 2026-06-17
arXiv:2606. 18183v1 Announce Type: new Abstract: Temporal difference (TD) learning with linear function approximation is a core method for policy evaluation.
Read next because A Diffusion Approximation for Temporal-Difference Learning with Linear Features under Markovian Noise overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: class, under, eval, line, project, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2606.18183v1 Announce Type: new Abstract: Temporal difference (TD) learning with linear function approximation is a core method for policy evaluation. Its classical continuous-time description is an ordinary differential equation (ODE), which captures the asymptotic mean dynamics but neglects stochastic fluctuations determining the error floor. We introduce a stochastic differential equation (SDE) approximation for linear TD(0) under Markovian noise. The resulting model distinguishes the contraction dynamics governed by the projected Bellman operator from the influence of Markovian sampling. As a consequence, the model explains the constant-stepsize error floor through the interaction between Markovian long-run covariance and the contraction geometry of the projected Bellman operator.
Potential threat/caveat for clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)": this item discusses evaluation.
source ↗open
score 100arxiv stat.ML (Machine Learning)arxiv:2606.18011unread
Fast Nonparametric Conditional Independence Testing via Two-Stage Regression
Eric V. Strobl · 2026-06-17
arXiv:2606. 18011v1 Announce Type: new Abstract: Constraint-based causal discovery relies on repeated conditional independence tests, but fast nonparametric tests often sacrifice calibration, especially when variables depend on the conditioning set through nonlinear relationships.
Read next because Fast Nonparametric Conditional Independence Testing via Two-Stage Regression overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "Add C2 control arm (donor sees marker_B without marker_A) to disambiguate paired-marker binding from marker_B leaking alone". Matching terms: under, line, rate, control, stage, test. Source: arxiv stat.ML (Machine Learning).
arXiv:2606.18011v1 Announce Type: new Abstract: Constraint-based causal discovery relies on repeated conditional independence tests, but fast nonparametric tests often sacrifice calibration, especially when variables depend on the conditioning set through nonlinear relationships. We introduce BLITZ (Broad-to-Local Independence Testing via residualiZation), a nonparametric conditional independence test designed to run well under a second while maintaining the accuracy needed for the thousands of queries performed by constraint-based causal discovery algorithms. BLITZ first removes broad smooth dependence on the conditioning set using low-order polynomial regression, then applies a small nonlinear feature map and residualizes those features with shallow tree regressions. The resulting statistic tests residual cross-covariance, with a moment-matched chi-square approximation to the null distribution. We show theoretically that the two-stage design reduces the effective complexity faced by the tree residualizers, allowing shallow trees to control residual conditional-mean bias while avoiding excessive overfitting. In simulations, BLITZ provides better null calibration than fast kernel, random-feature, and regression-based competitors while remaining among the fastest methods tested. In causal discovery experiments on synthetic graphs and flow-cytometry data, BLITZ yields more reliable endpoint orientations among retained adjacencies and competitive structural recovery. These results suggest that broad-to-local residualization is a practical route to calibrated, scalable nonparametric conditional independence testing for causal discovery.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses bias.
source ↗open
score 100arxiv stat.ML (Machine Learning)arxiv:2606.17684unread
Geometrical fairness in graph neural networks
Arturo P\'erez-Peralta, Sandra Ben\'itez-Pe\~na, Blas Kolic, Rosa E. Lillo · 2026-06-17
arXiv:2606. 17684v1 Announce Type: new Abstract: Graph-based learning methods have become increasingly prominent due to their strong performance across diverse applications.
Read next because Geometrical fairness in graph neural networks overlaps with clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, under, eval, rate, project, propagate, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2606.17684v1 Announce Type: new Abstract: Graph-based learning methods have become increasingly prominent due to their strong performance across diverse applications. Among these, recent frameworks grounded in diffusion processes provide a unifying perspective that extends traditional graph neural network formulations while addressing limitations of standard message-passing mechanisms. Despite these advances, concerns remain regarding the fairness of such models, as they may propagate or amplify biases present in the data. In this work, we introduce a fairness-aware adaptation of graph-based diffusion by modifying the underlying Laplacian operator. Our approach incorporates multiple complementary transformations, including subspace projections, spectral adjustments, and frequency-based filtering, to mitigate bias-related components. Leveraging the intrinsic smoothing properties of graph diffusion, we provide a principled analysis of the resulting behavior and establish theoretical insights into fairness properties. We evaluate the proposed framework on both synthetic and real-world datasets, demonstrating that it achieves competitive performance while improving fairness metrics with limited additional computational cost.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits at 23.5% when a co-trained partner learns <A>..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses limitation, limitations, bias.
source ↗open
score 100arxiv stat.ML (Machine Learning)arxiv:2606.17426unread
Bounded Difference Concentration for Infinitely Exchangeable Sequences with Applications to AI Benchmark Uncertainty
Fangyuan Lin, Spencer Frei, Victor H. de la Pena · 2026-06-17
arXiv:2606. 17426v1 Announce Type: new Abstract: We consider the concentration properties of functions of infinitely exchangeable random variables.
Read next because Bounded Difference Concentration for Infinitely Exchangeable Sequences with Applications to AI Benchmark Uncertainty overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: rect, line, rate, full, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2606.17426v1 Announce Type: new Abstract: We consider the concentration properties of functions of infinitely exchangeable random variables. By conditioning on the de Finetti directing measure, we show that the deviation of any function with bounded-difference constants $c_1, \dots, c_n$ decomposes into a conditional sampling fluctuation and a latent mixture fluctuation. When this latent mixture is $\sigma_{\mathrm{mix}}^2$-subgaussian, we establish a concentration inequality with an effective variance proxy of $\frac{1}{4}\sum_i c_i^2 + \sigma_{\mathrm{mix}}^2$. Crucially, we demonstrate that for zero-sum linear contrasts, such as the difference between a subsample mean and a full population mean, the latent mixture term cancels exactly. This cancellation yields a tight, mixture-free Hoeffding-type bound that provides a direct de Finetti mechanism for the infinite-extendibility limit of recent finite-exchangeable concentration results. We apply this framework to quantify uncertainty in composite AI benchmarks, such as MMLU, where question items naturally exhibit exchangeable dependence across domains. Our results provide both a domain-stratified hierarchical model for bounding the uncertainty of accuracy scores, and a distribution-free, cost-saving statistical guarantee for accurately estimating full benchmark scores from random subsets.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses benchmark.
source ↗open
score 92arxiv cs.CR (Cryptography and Security)arxiv:2606.17245unread
Cache to the Future: A Distributed Webpage Archive for Internet Blackouts
Ross Evans, Diogo Barradas · 2026-06-17
arXiv:2606. 17245v1 Announce Type: new Abstract: Internet blackouts, occurring due to technological mishaps or intentional governmental action, prevent citizens from accessing the internet.
Read next because Cache to the Future: A Distributed Webpage Archive for Internet Blackouts overlaps with clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "#351 follow-up: broader-vocab position-0 sweep at T=1.0 + position-1 suffix isolation". Matching terms: source, rate. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.17245v1 Announce Type: new Abstract: Internet blackouts, occurring due to technological mishaps or intentional governmental action, prevent citizens from accessing the internet. Citizens in regions where internet blackouts are common have utilized blackout-resistant technologies to maintain communication. Such technologies often rely on mobile mesh networks to provide limited messaging services. However, no technology currently exists which can provide continued access to knowledge sources on the web during a blackout. We present Cache to the Future (CttF): a system to cache and deliver static content hosted on the web during a blackout. CttF's distributed community ratings crowdsources caching at scale while cryptographic constructs (digital signatures, proofs-of-work) mitigate adversarial interference. Our realistic simulations demonstrate CttF delivering content at city-scale across a wide range of benign and adversarial scenarios.
Potential threat/caveat for clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)": this item discusses adversarial.
source ↗open
score 84arxiv cs.CR (Cryptography and Security)arxiv:2606.17223unread
Safety, Security, and Cognitive Risks in Neuro-Symbolic AI
Manoj Parmar · 2026-06-17
arXiv:2606. 17223v1 Announce Type: new Abstract: Neuro-symbolic AI (NeSy) pairs neural perception with symbolic reasoning, making it attractive for high-stakes domains where explainability and structured inference are required.
Read next because Safety, Security, and Cognitive Risks in Neuro-Symbolic AI overlaps with clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: line, model, absent. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2606.17223v1 Announce Type: new Abstract: Neuro-symbolic AI (NeSy) pairs neural perception with symbolic reasoning, making it attractive for high-stakes domains where explainability and structured inference are required. However, this hybrid architecture introduces an enlarged attack surface spanning five layers: neural perception, symbolic knowledge bases, reasoning engines, agentic orchestration, and data stores -- each exploitable in ways absent from purely neural systems. This paper makes six contributions: (1) formal definitions of NeSy Attack Surface, Symbolic Integrity Violation (SIV), and Cross-Layer Amplification Ratio $\mathcal{X}$, decomposed into neural-caused and autonomous symbolic sensitivity components; (2) a unified threat model extending MITRE ATLAS with 11 NeSy-specific tactic extensions and a five-profile attacker taxonomy; (3) a symbolic-layer threat catalogue covering knowledge graph (KG) poisoning, ontology-merging, and inference-engine subversion; (4) analysis of cognitive risks -- automation bias, authority bias, and sycophantic reinforcement -- structurally amplified by NeSy's explicit logical explanations relative to black-box neural outputs; (5) interdisciplinary mitigations with measurable acceptance criteria aligned to NIST AI 600-1 and the EU AI Act; (6) three empirical benchmarks: (E1) targeted KG poisoning achieves break-even SIV at injection budget $B=5$ on a 205-entity medical KG, with a KG-specific stealth/SIV trade-off; (E2) PGD-10 at $\varepsilon=0.01$ yields $\mathcal{X}=5.884$ (95% CI $[4.64,\, 8.00]$, $p<0.0001$), confirmed adversarially specific by a matched-random baseline ($E^{R}_{\mathrm{rand}}=0$), on a DistilBERT+ProbLog pipeline; (E3) single-axiom OWL edits achieve 93.3% SIV success with 100% Pellet-consistency stealth, but held-out STIX detection fails at 50% (random-guessing level), an open problem.
Potential threat/caveat for clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)": this item discusses bias, adversarial, benchmark.
source ↗open
score 64M7 QA inline RSS threat sourceunread
Artifact verification caveats for Sagan clean results
M7 QA · No release date
This paper studies failure modes and caveats when Sagan creates a clean result only after verifying an artifact row. It proposes benchmark checks for artifact verification, clean-result review comments, and negative controls.
Read next because Artifact verification caveats for Sagan clean results overlaps with experiment "Add C2 control arm (donor sees marker_B without marker_A) to disambiguate paired-marker binding from marker_B leaking alone". Matching terms: control. Source: M7 QA inline RSS threat source.
This paper studies failure modes and caveats when Sagan creates a clean result only after verifying an artifact row. It proposes benchmark checks for artifact verification, clean-result review comments, and negative controls.
Potential threat/caveat for experiment "Add C2 control arm (donor sees marker_B without marker_A) to disambiguate paired-marker binding from marker_B leaking alone": this item discusses failure, caveat, caveats, negative, benchmark.
source ↗open