All papers

Identifies a specific vision transformer attention head that spuriously attends to medical image corner text instead of lung pathology.

Proposes an availability framework to identify and resolve within-event timing mismatches in earnings-call financial NLP benchmarks.

Improves reinforcement learning credit assignment in multimodal reasoning using pairwise token contrast and adaptive trajectory confidence.

Frames diffusion language models as open quantum systems to enable text generation conditioned on superpositions of prompts and thoughts.

Introduces a system that generates and compares alternative scene drafts to guide full dramatic script adaptations from prose.

Demonstrates that machine unlearning often fails to prevent targeted information recovery across alternative multilingual access paths.

Analyzes how specific nouns and semantic clusters relate to answer selection and user engagement in logistics question-answering forums.

Introduces a non-autoregressive transformer model that dynamically decides tag generation order at inference time for low-resource sequence labeling.

Presents a training-free framework that tracks changing conversational facts as versioned records to improve agent memory retrieval.

Proposes a reinforcement learning credit-assignment method for document parsing models to align rewards with structured page elements.

Introduces a benchmark that structures multi-pull-request code changes into dependency graphs to evaluate long-horizon software engineering agents.

Evaluates whether internal hidden-state probes can identify failure modes and pinpoint failing steps in multi-step reasoning traces.

Measures demographic bias related to age, gender, and race based on first names across local language models.

Applies a dual language model consensus pipeline to extract clinical phenotypes from electronic health records for patients with Rett syndrome.

Presents a benchmark that measures long-context reasoning difficulty using evidence scope, dispersion, and reasoning depth.

Specializes a translation model to a specific indigenous Nawatl variety using a curated dataset and finite-state morphological analysis.

Introduces a benchmark to evaluate whether text-to-SQL systems can detect flawed query premises and abstain from generating invalid SQL.

Develops a two-stage training approach combining evidence reasoning and reinforcement learning to improve pragmatic social reasoning in conversational models.

Proposes an event-centric framework to align audio and video features when generating detailed video descriptions with language models.

Introduces a compact language model trained with distillation and reinforcement learning to manage long-term memory across multiple external frameworks.

Analyzes the overuse and contextual appropriateness of a common corrective phrasing pattern in Mandarin language model text.

Presents an adversarial training approach designed to improve language model honesty across diverse benchmarks without degrading core capabilities.

Uses diffusion language models to perform multi-target few-shot infilling across shared code context guided by demonstration examples.

Evaluates the ability of language models to predict state changes in discrete spatial grids across sequential operations.

Presents an evaluation framework combining simulated mental health patients and expert-developed rubrics to assess conversational language models.

Introduces a set-level reinforcement learning objective to train language models to produce diverse and document-grounded scientific research ideas.

Proposes a visual verifier that decomposes complex prompts into sub-questions and assigns continuous scores to improve multimodal evaluation.

Tests whether transcripts of correctly solved tasks from saturated benchmarks can predict model performance on unseen harder evaluations.

The paper introduces a modular measurement system to evaluate how agent frameworks, tools, and memory impact verifiable reasoning tasks.

The paper introduces a parameter decomposition method to interpret the mechanistic subcomponents of language model weights.

The paper identifies an internal activation direction in spoken language models to control and transfer conversational turn-taking styles.

The paper presents a metric and benchmark measuring how quickly environment dynamics can change before agent decision-making fails.

The paper presents a framework and dataset for learning to detect and repair errors during multi-agent system execution.

The paper analyzes how answering via a separate model call improves evidence adherence when reasoning traces conflict with provided context.

The paper presents an agent that expands its domain-specific language for exact answering without external rewards or judges.

The paper trains reasoning models to perform sequential experimental design by optimizing information-gain rewards during action selection.

The paper investigates performance vulnerabilities in self-improving agents when their co-evolved skill banks undergo post-deployment drift.

The paper analyzes how vision-language models process negation by comparing model attention mechanisms with human eye-tracking data.

The paper presents a self-play reinforcement learning framework to simulate interactions between vehicles, pedestrians, and cyclists without human demonstrations.

The paper analyzes why reinforcement learning with majority-vote rewards provides effective training gradients despite noisy individual labels.

The paper presents a zero-shot vision-language method for generating task-oriented robotic grasps in 6 degrees of freedom.

The paper develops replicable algorithms with sublinear regret bounds for the dueling bandits problem.

The paper formalizes compositional structure in complex systems using algorithmic information theory and compression concepts.

The paper proposes a non-elitist evolutionary optimization framework that retains and iteratively refines low-reward agent skills using execution feedback.

The paper studies how semantic layers, orchestration, and compaction impact the performance and behavior of data agents across benchmarks.

The paper investigates when language models learn capabilities by tracking unembedding matrix dynamics during pretraining using parameter-trajectory crosscoding.

Presents a multi-agent system for paper screening and data extraction in systematic literature reviews following standard guidelines.

Combines language models and lightweight decision models to fact-check statements in peer reviews against original manuscripts.

Develops an open agent environment for evaluating how autonomously language agents can conduct post-training machine learning experiments.

Applies trajectory Bellman residual minimization to turn sparse rewards into dense token signals for arithmetic reasoning.

Estimates language model ensemble member reliability from unlabeled agreement patterns using an expectation-maximization jury formulation.

Introduces an equivariant self-supervised learning framework that avoids topological collapse in 3D point cloud masked autoencoding.

Evaluates test-time search in computer-use agents, finding critics perform better when invoked only at predicted final states.

Uses reinforcement learning to train an agent module that selects, sizes, and orders retrieved memories for downstream tasks.

Measures tool-agent performance failures under controlled harness limits such as completion caps and tool menu visibility.

Analyzes trade-offs between aggressive and conservative planning horizons in language models operating under partial observability.

Introduces a dataset benchmarking human-agent interaction and collaboration across long-horizon professional workflows.

Presents a decomposed network architecture that applies separate graph and convolution modules to trend and seasonal time series components.

Examines failure modes in label-free evaluation where models reach high consistency across perturbations while generating identical incorrect answers.

Analyzes working memory error patterns in language models to show that failures stem from readout interference rather than target retrieval.