All papers

Presents a benchmark with human-annotated visual evidence to evaluate retrieval and question answering across long documents.

Deconstructs multiple-choice option bias in language models into two factors tracking answer flips and positional skew.

Explores methods to extract the underlying decisive sub-questions that drive language model decisions across diverse generated outputs.

Proposes an adaptive reasoning framework that supervises intermediate retrieval states and stops reasoning steps dynamically to cut latency.

Introduces a benchmark to test model generalization under systematic convenience sampling biases in tabular data.

The paper proposes SHIP, an architecture for extracting structured differentially private representations from text for analytical queries.

The paper examines AI-generated text detection in scientific papers by combining visual formatting, PDF text, and LaTeX representations.

The paper proposes RoPERelay to reorganize positional geometry and rotary embeddings after sparse KV cache token eviction.

The paper introduces a causal intervention framework to distill transferable self-generated in-context learning demonstrations.

The paper presents an automated speech recognition and sentence completion system to assist dementia patients speaking Bengali.

The paper surveys memory-centric compression techniques for managing inference resources and context windows in large language models.

The paper introduces ExecutionRAG to show how search drift causes apparent retrieval failures in multi-hop question answering.

The paper introduces the Korean Benchmark Suite to evaluate foundation models across six natural language capabilities.

The paper audits data decontamination methods on math corpora to quantify false removals and benchmark leakage risks.

The paper presents a teacher-student framework using language models for weak annotation to classify cybersecurity tweets.

The paper proposes isolating internal model judgments from instruction formatting variations using a label-free hidden state subspace.

The paper introduces TimeCap, an open deterministic pipeline for generating signal-grounded captions from wearable health time-series data.

The paper evaluates output non-determinism in commercial language model deployments during medical device regulatory compliance assessments.

The paper introduces VALCON to evaluate and improve value consistency when language models alternate between generator and evaluator roles.

The paper presents PocketTSLM, deploying an on-device soft-prompt time-series language model on smartphones for personal health data.

The paper introduces ZoomCap, an adaptive visual zooming and evidence-routing framework to improve fine-grained image captioning supervision.

The paper introduces a hybrid generative-tagging architecture to improve efficiency and coverage in product attribute and value extraction.

The paper presents VoteGEO-Bench to measure how rewriting web content influences information visibility in generative search engine election queries.

The paper introduces QuasiLeak-Bench to evaluate whether privacy detectors can identify distributed quasi-identifiers governed by disclosure policies.

The paper studies ranking and partitioning attention heads by gradient norm for sparse and continual reinforcement learning fine-tuning.

The paper presents an internal attention boosting method to improve how language models infer implicit conversational affective states.

The paper introduces RenoLayout to generate residential renovation floorplans and furniture layouts under structural and physical constraints.

Introduces a dynamic precision control framework that allocates weight chunks to mixture-of-experts models during inference.

Analyzes how calibration language choices affect block-influence depth pruning rankings in multilingual language models.

Proposes a factor-wise representation framework to selectively treat different distribution shifts in multimodal machine learning.

Evaluates prompt optimization methods and identifies effective prompting strategies for social media stress and emotion classification.

Introduces a hybrid speculative decoding framework splitting draft tasks between mobile neural processing units and cloud servers.

Develops a lightweight situational distortion metric to evaluate speech recognition errors based on practical conversational severity.

Identifies translation circuits in multilingual models and optimizes fewer than five percent of parameters for low-resource translation.

Conducts a controlled evaluation of direct knowledge injection into addressed conditional memory modules across small and production models.

Gates reinforcement learning policy-shift distillation using intermediate checkpoints to ensure student models only learn consistent teacher shifts.

Introduces a data augmentation method that breaks spurious contextual correlations causing emergent misalignment during domain finetuning.

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.

Shows across 200 models that proposed intrinsic tokenizer metrics do not reliably predict downstream language model performance.

Reviews and formalizes data selection, weighting, and scheduling policies across the pretraining and post-training lifecycle of language models.

Compares abstract-only versus full-text inputs and few-shot versus fine-tuning approaches for generating biomedical lay summaries.

Applies adversarial training and distillation to remove unwanted gender bias in audio-enabled language models.

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.

Proposes a compressed representation using Gaussian trellises to quantize conditional memory lookup tables down to two bits.

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

Distills multiple speech attributes from four teacher models into a single shared speech encoder.

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

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.

Studies how using UTF-16 byte order improves boundary prediction and representation efficiency in multilingual byte-level language models.

Introduces a compression-based patching strategy for byte-level language models that allocates compute where upcoming bytes are unlikely to repeat.

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.