MI-CXR is a new benchmark that shows state-of-the-art vision-language models achieve only 29.3% accuracy on longitudinal reasoning tasks across multi-visit chest X-ray sequences.
Mixed citations
Title resolution pending
Mixed citation behavior. Most common role is background (50%).
117
Pith papers citing it
Background
50% of classified citations
citation-role summary
background 3
method 3
citation-polarity summary
representative citing papers
Pre-training 6B LLMs on temporally ordered Common Crawl snapshots yields models with improved factual freshness and temporal precision over shuffled baselines while matching on general language understanding.
Life-Harness evolves reusable interventions from training trajectories to enhance frozen LLM agents on unseen tasks across seven deterministic environments, yielding 88.5% average relative improvement in 116 of 126 model-environment settings.
A new 263k TTCW-annotated story dataset shows non-reasoning fine-tuning of Qwen3 models outperforms reasoning-supervised fine-tuning for fixed-format long-form literary review generation.
RISE is an inference-time semantic reranking framework that refines low-confidence predictions in rhetorical role labeling using contrastively learned label representations, delivering an average +9.15 macro-F1 gain on hard examples across eight datasets and seven models.
Introduces the UCSF-PDGM-VQA dataset of 2387 QA pairs from 473 glioma MRI studies and demonstrates that state-of-the-art VLMs exhibit modality collapse on multi-sequence 3D medical images.
Rule2DRC is a benchmark for LLM agents synthesizing DRC scripts from natural language rules, paired with SplitTester that improves Best-of-N selection via execution-guided discriminative test generation.
Proves that RoPE attention loses locality bias and token distinction in long contexts, approaching random behavior independent of content.
QAOD projects away question-aligned directions from answer representations to isolate domain-agnostic factuality signals, enabling efficient hallucination detection with top in-domain AUROC and up to 21% better OOD transfer.
DiHAL uses geometry proxies to pick where to replace the lower layers of a pretrained transformer with a diffusion bridge for hidden-state reconstruction, improving over token-level diffusion baselines on 8B models.
Phoenix-bench shows agentic AI systems lose 37-58% resolved rate when moving from SWE-bench Verified to hardware tasks because bugs spread across parallel modules via signal flow, with testbench feedback lifting performance by 42-45% while file-level oracles add only 1.4%.
TokAlign++ learns token alignments between LLM vocabularies from monolingual representations to enable faster adaptation, better text compression, and effective token-level distillation across 15 languages with minimal steps.
A parallel multi-turn medical dialogue dataset spanning English and nine Indic languages is created from synthetic consultations to enable personalized AI healthcare interactions.
AcquisitionSynthesis uses acquisition functions as rewards to train generators that produce higher-quality synthetic data, delivering 2-7% gains on math, medical QA, and coding tasks with improved robustness to forgetting.
OLIVIA treats LLM agent action selection as a contextual linear bandit over frozen hidden states and applies UCB exploration to adapt online, yielding consistent gains over static ReAct and prompt-based baselines on four benchmarks.
MMM-Bench supplies 5,990 multi-modal documents from 12 commercial domains annotated along a 5-level taxonomy to test document classification under realistic business conditions.
Fully Looped Transformer stabilizes looped training up to 12 iterations via distributed inter-loop signals and attention injection, improving downstream performance by up to 13.2%.
ALiBi bias is the expectation of positional LSH-induced block masks, yielding spectral and max-norm approximation bounds that reduce long-context biased attention to randomized short-context unbiased attention.
LLMs copy biased analyst ratings in investment decisions but a new detection method encourages independent reasoning and can improve stock return predictions beyond human levels.
Agentick is a new benchmark for sequential decision-making agents that evaluates RL, LLM, VLM, hybrid, and human approaches across 37 tasks and finds no single method dominates.
VITA-QinYu is the first expressive end-to-end spoken language model supporting role-playing and singing alongside conversation, trained on 15.8K hours of data and outperforming prior models on expressiveness and conversational benchmarks.
Post-Reasoning boosts LLM accuracy by reversing the usual answer-after-reasoning order, delivering mean relative gains of 17.37% across 117 model-benchmark pairs with zero extra cost.
Finite-answer projections of continuation probabilities stabilize before the answer is parseable, showing 17-31 token mean lead in delayed-verdict tasks with Qwen3-4B-Instruct.
Prepending stochastic sequences from Lorem Ipsum vocabulary to prompts during GRPO resampling broadens reasoning exploration and outperforms standard resampling on hard tasks for 1.7B-7B models.
citing papers explorer
-
MI-CXR: A Benchmark for Longitudinal Reasoning over Multi-Interval Chest X-rays
MI-CXR is a new benchmark that shows state-of-the-art vision-language models achieve only 29.3% accuracy on longitudinal reasoning tasks across multi-visit chest X-ray sequences.
-
Understanding Data Temporality Impact on Large Language Models Pre-training
Pre-training 6B LLMs on temporally ordered Common Crawl snapshots yields models with improved factual freshness and temporal precision over shuffled baselines while matching on general language understanding.
-
Adapting the Interface, Not the Model: Runtime Harness Adaptation for Deterministic LLM Agents
Life-Harness evolves reusable interventions from training trajectories to enhance frozen LLM agents on unseen tasks across seven deterministic environments, yielding 88.5% average relative improvement in 116 of 126 model-environment settings.
-
When Reasoning Supervision Hurts: TTCW-Based Long-Form Literary Review Generation
A new 263k TTCW-annotated story dataset shows non-reasoning fine-tuning of Qwen3 models outperforms reasoning-supervised fine-tuning for fixed-format long-form literary review generation.
-
Semantic Reranking at Inference Time for Hard Examples in Rhetorical Role Labeling
RISE is an inference-time semantic reranking framework that refines low-confidence predictions in rhetorical role labeling using contrastively learned label representations, delivering an average +9.15 macro-F1 gain on hard examples across eight datasets and seven models.
-
UCSF-PDGM-VQA: Visual Question Answering dataset for brain tumor MRI interpretation
Introduces the UCSF-PDGM-VQA dataset of 2387 QA pairs from 473 glioma MRI studies and demonstrates that state-of-the-art VLMs exhibit modality collapse on multi-sequence 3D medical images.
-
Rule2DRC: Benchmarking LLM Agents for DRC Script Synthesis with Execution-Guided Test Generation
Rule2DRC is a benchmark for LLM agents synthesizing DRC scripts from natural language rules, paired with SplitTester that improves Best-of-N selection via execution-guided discriminative test generation.
-
RoPE Distinguishes Neither Positions Nor Tokens in Long Contexts, Provably
Proves that RoPE attention loses locality bias and token distinction in long contexts, approaching random behavior independent of content.
-
When Answers Stray from Questions: Hallucination Detection via Question-Answer Orthogonal Decomposition
QAOD projects away question-aligned directions from answer representations to isolate domain-agnostic factuality signals, enabling efficient hallucination detection with top in-domain AUROC and up to 21% better OOD transfer.
-
Where Should Diffusion Enter a Language Model? Geometry-Guided Hidden-State Replacement
DiHAL uses geometry proxies to pick where to replace the lower layers of a pretrained transformer with a diffusion bridge for hidden-state reconstruction, improving over token-level diffusion baselines on 8B models.
-
Is Agentic AI Ready for Real-World Hardware Engineering? A Deep Dive with Phoenix-bench
Phoenix-bench shows agentic AI systems lose 37-58% resolved rate when moving from SWE-bench Verified to hardware tasks because bugs spread across parallel modules via signal flow, with testbench feedback lifting performance by 42-45% while file-level oracles add only 1.4%.
-
TokAlign++: Advancing Vocabulary Adaptation via Better Token Alignment
TokAlign++ learns token alignments between LLM vocabularies from monolingual representations to enable faster adaptation, better text compression, and effective token-level distillation across 15 languages with minimal steps.
-
IndicMedDialog: A Parallel Multi-Turn Medical Dialogue Dataset for Accessible Healthcare in Indic Languages
A parallel multi-turn medical dialogue dataset spanning English and nine Indic languages is created from synthetic consultations to enable personalized AI healthcare interactions.
-
AcquisitionSynthesis: Targeted Data Generation using Acquisition Functions
AcquisitionSynthesis uses acquisition functions as rewards to train generators that produce higher-quality synthetic data, delivering 2-7% gains on math, medical QA, and coding tasks with improved robustness to forgetting.
-
OLIVIA: Online Learning via Inference-time Action Adaptation for Decision Making in LLM ReAct Agents
OLIVIA treats LLM agent action selection as a contextual linear bandit over frozen hidden states and applies UCB exploration to adapt online, yielding consistent gains over static ReAct and prompt-based baselines on four benchmarks.
-
Multi-domain Multi-modal Document Classification Benchmark with a Multi-level Taxonomy
MMM-Bench supplies 5,990 multi-modal documents from 12 commercial domains annotated along a 5-level taxonomy to test document classification under realistic business conditions.
-
Simply Stabilizing the Loop via Fully Looped Transformer
Fully Looped Transformer stabilizes looped training up to 12 iterations via distributed inter-loop signals and attention injection, improving downstream performance by up to 13.2%.
-
Positional LSH: Binary Block Matrix Approximation for Attention with Linear Biases
ALiBi bias is the expectation of positional LSH-induced block masks, yielding spectral and max-norm approximation bounds that reduce long-context biased attention to randomized short-context unbiased attention.
-
Fin-Bias: Comprehensive Evaluation for LLM Decision-Making under human bias in Finance Domain
LLMs copy biased analyst ratings in investment decisions but a new detection method encourages independent reasoning and can improve stock return predictions beyond human levels.
-
Agentick: A Unified Benchmark for General Sequential Decision-Making Agents
Agentick is a new benchmark for sequential decision-making agents that evaluates RL, LLM, VLM, hybrid, and human approaches across 37 tasks and finds no single method dominates.
-
VITA-QinYu: Expressive Spoken Language Model for Role-Playing and Singing
VITA-QinYu is the first expressive end-to-end spoken language model supporting role-playing and singing alongside conversation, trained on 15.8K hours of data and outperforming prior models on expressiveness and conversational benchmarks.
-
Post Reasoning: Improving the Performance of Non-Thinking Models at No Cost
Post-Reasoning boosts LLM accuracy by reversing the usual answer-after-reasoning order, delivering mean relative gains of 17.37% across 117 model-benchmark pairs with zero extra cost.
-
When Does a Language Model Commit? A Finite-Answer Theory of Pre-Verbalization Commitment
Finite-answer projections of continuation probabilities stabilize before the answer is parseable, showing 17-31 token mean lead in delayed-verdict tasks with Qwen3-4B-Instruct.
-
Nonsense Helps: Prompt Space Perturbation Broadens Reasoning Exploration
Prepending stochastic sequences from Lorem Ipsum vocabulary to prompts during GRPO resampling broadens reasoning exploration and outperforms standard resampling on hard tasks for 1.7B-7B models.
-
Statistical Consistency and Generalization of Contrastive Representation Learning
The paper proves statistical consistency of contrastive loss to optimal ranking via an AUC criterion and derives generalization bounds O(1/m + 1/sqrt(n)) for supervised and O(1/sqrt(m) + 1/sqrt(n)) for self-supervised CRL that explain benefits of large negative sets.
-
LIMSSR: LLM-Driven Sequence-to-Score Reasoning under Training-Time Incomplete Multimodal Observations
LIMSSR reformulates incomplete multimodal learning as LLM-driven sequence-to-score reasoning with prompt-guided imputation and mask-aware aggregation, outperforming baselines on action quality assessment without complete training data.
-
BWLA: Breaking the Barrier of W1AX Post-Training Quantization for LLMs
BWLA is the first post-training quantization method for LLMs that achieves 1-bit weights paired with low-bit activations such as 6 bits, using OKT to reshape weights and suppress activation tails plus PSP for low-rank refinement.
-
Revisiting Non-Verbatim Memorization in Large Language Models: The Role of Entity Surface Forms
LLMs display inconsistent factual recall across different surface forms of the same entity, with greater robustness to minor spelling changes than to aliases or abbreviations.
-
PaperMind: Benchmarking Agentic Reasoning and Critique over Scientific Papers in Multimodal LLMs
PaperMind is a new benchmark that evaluates integrated multimodal reasoning and critique over scientific papers through four complementary task families across seven domains.
-
STAR-Teaming: A Strategy-Response Multiplex Network Approach to Automated LLM Red Teaming
STAR-Teaming uses a Strategy-Response Multiplex Network inside a multi-agent framework to organize attack strategies into semantic communities, delivering higher attack success rates on LLMs at lower computational cost than prior methods.
-
Dual-View Training for Instruction-Following Information Retrieval
Dual-view training via polarity reversal improves instruction-following retrieval performance by 45% on the FollowIR benchmark using a 305M-parameter encoder.
-
Evolutionary Negative Module Pruning for Better LoRA Merging
ENMP prunes negative LoRA modules via evolutionary search to boost merging performance to new state-of-the-art levels across language and vision tasks.
-
HORIZON: A Benchmark for In-the-wild User Behaviour Modeling
HORIZON creates a cross-domain, long-horizon user modeling benchmark from Amazon Reviews that tests generalization across time, domains, and unseen users, exposing gaps in sequential and LLM-based recommendation models.
-
WhatsApp Vaccine Discourse (WhaVax): An Expert-Annotated Dataset and Benchmark for Health Misinformation Detection
WhaVax is a new expert-annotated dataset of WhatsApp vaccine messages with benchmarks showing competitive performance from embeddings and LLMs for misinformation detection under data scarcity.
-
SWE-Mutation: Can LLMs Generate Reliable Test Suites in Software Engineering?
SWE-Mutation benchmark shows current LLMs achieve low verification (10.20%) and detection (36.15%) rates on 2,636 mutated variants, exposing weaknesses in generating reliable test suites.
-
How Far Will They Go? Red-Teaming Online Influence with Large Language Models
An empirical red-teaming study measures political Overton Windows across more than 30 open-source LLMs from 10 families and finds left-leaning bias, inverse size correlation, regional variation, and variable jailbreak effectiveness.
-
DelTA: Discriminative Token Credit Assignment for Reinforcement Learning from Verifiable Rewards
DelTA estimates token coefficients to amplify discriminative directions in token-gradient vectors, reweighting the RLVR surrogate to produce more contrastive side-wise centroids and yielding 3.26 and 2.62 point gains on math benchmarks for 8B and 14B Qwen3 models.
-
The Illusion of Intervention: Your LLM-Simulated Experiment is an Observational Study
Interventions in LLM-simulated user experiments induce distribution shifts in latent attributes that create confounding bias, diagnosable with negative control outcomes and partially mitigated by adding setting-relevant persona details.
-
GoLongRL: Capability-Oriented Long Context Reinforcement Learning with Multitask Alignment
GoLongRL releases a 23K-sample open long-context RL dataset spanning 9 tasks and introduces TMN-Reweight to improve multitask optimization, achieving performance comparable to much larger models under GRPO.
-
Weasel: Out-of-Domain Generalization for Web Agents via Importance-Diversity Data Selection
Weasel is a trajectory selection method that improves out-of-domain generalization for web agents while achieving 9.7-12.5x training speedups via importance-diversity optimization, AXTree pruning, and rationale style matching.
-
A Multi-Agent Framework for Feature-Constrained Difficulty Control in Reading Comprehension Item Generation
MAFIG is a multi-agent framework that uses LLM agents and evaluators to generate reading comprehension items with significantly higher adherence to specified feature constraints than single-agent baselines.
-
Forecasting Downstream Performance of LLMs With Proxy Metrics
Proxy metrics from next-token distributions over expert solutions outperform loss and compute baselines for ranking LLMs, selecting pretraining data, and extrapolating performance across compute scales.
-
AMARIS: A Memory-Augmented Rubric Improvement System for Rubric-Based Reinforcement Learning
AMARIS augments rubric updates in RL for LLMs with a persistent memory of rollout analyses and prior edits, yielding gains such as +2.8 points on GPQA-Diamond over local-adaptive baselines.
-
iPOE: Interpretable Prompt Optimization via Explanations
iPOE generates and optimizes annotation guidelines from explanations to produce interpretable prompts, reporting up to 39% gains over baselines on four datasets with LLM explanations substituting for human ones.
-
Predictive Prefetching for Retrieval-Augmented Generation
Introduces predictive prefetching for RAG that anticipates retrieval needs several tokens ahead via three components, reporting up to 43.5% latency reduction and 62.4% TTFT improvement while preserving answer quality.
-
AutoVecCoder: Teaching LLMs to Generate Explicitly Vectorized Code
AutoVecCoder combines VecPrompt for automated intrinsic knowledge synthesis and VecRL for efficiency-aligned RL to train an 8B LLM that achieves SOTA on SimdBench SSE/AVX subsets and sometimes exceeds -O3 compiler results.
-
Generalization or Memorization? Brittleness Testing for Chess-Trained Language Models
A compact 25M chess move predictor exceeds larger fine-tuned models on puzzles, indicating memorization in earlier claims, while LLM-Modulo raises general LLM move accuracy from 1.2% to 21.2% and validity to 95.3%.
-
Prefix-Adaptive Block Diffusion for Efficient Document Recognition
PA-BDM adapts block diffusion by switching to causal intra-block denoising and dynamically committing reliable prefixes to KV cache, yielding higher accuracy and 71.6% higher throughput than a comparable baseline on document benchmarks.
-
GRLO: Towards Generalizable Reinforcement Learning in Open-Ended Environments from Zero
GRLO shows RLHF from scratch on 5K open-ended prompts raises average performance from 24.1 to 63.1 across domains on Qwen3-4B-Base using 46x less data and 68x less compute than in-domain RLVR while remaining competitive with heavily post-trained models.
-
Pause and Reflect: Conformal Aggregation for Chain-of-Thought Reasoning
A conformal procedure for CoT replaces majority voting with weighted aggregation and calibrates abstention to guarantee low confident-error rates, achieving 90.1% selective accuracy on GSM8K by abstaining on under 5% of cases.