All papers

The paper tests activation steering as a text watermarking technique, finding that standard logit-based watermarking provides more robust detection against paraphrasing.

The paper proposes a training approach that uses learned chunk routing to speed up long-context diffusion language model processing.

The paper introduces Swiss4Arg, a multilingual argument mining dataset based on official Swiss voting brochures across four national languages.

The paper evaluates whether providing page images to vision-language models improves the detection of errors in Japanese-English manga translation.

The paper proposes Batch-Paced, a method that dynamically allocates decoding compute across batched requests to improve discrete diffusion language model inference.

The paper introduces Agent-Reflex, a training framework designed to teach language agents how to reflect and recover from noisy environment feedback.

The paper introduces Fish Audio Drama 3, an open-source text-to-speech model designed for multi-character dialogue, long-form synthesis, and iterative voice revision.

The paper examines machine grading on undergraduate essays across three universities, finding models compress score distributions and vary across institutions.

The paper introduces BanglaTOD, a tri-lingual benchmark for task-oriented dialogue understanding covering Bangla, English, and romanized Banglish across nine service domains.

The paper introduces an optimization method that jointly refines communication instructions between communicating agents to improve multi-agent collaboration.

The paper probes language model hidden states across generated stories to test whether models anticipate narrative events multiple steps ahead.

The paper evaluates typed decision models against standard language model judges to measure how emotional tone and biases affect their evaluation accuracy.

The paper presents SparPR, a runtime key-value cache recovery mechanism that reduces memory bottlenecks during model inference on consumer hardware.

Evaluates AI text detectors under attacks applied to both human and machine text, revealing high false-positive rates on edited human writing.

Proposes dynamic activation steering using preceding denoising predictions to reduce hallucinations in diffusion language models.

Presents a pragmatic prompting framework to preserve speaker voice, emotion, and tone during speech-to-speech translation.

Surveys methods for compressing large language model benchmarks to reduce evaluation costs while maintaining accurate score estimates.

Presents a multilingual framework and dataset spanning four languages to evaluate clinical pain representations across exams and clinical notes.

Introduces a human-verified benchmark to evaluate how accurately reward models judge language model outputs on legal tasks.

Develops a simulation framework using persona-guided models to optimize comparative doctor summary presentations for medical patients.

Introduces a low-latency guardrail model that routes incoming agent actions through editable policy trees for safety auditing.

Probes layer representations in multilingual language models to study whether internal color geometry aligns with human perception or language syntax.

Demonstrates an attack method that alters embedded multi-bit text watermarks without knowing the secret generation key.

Equips diffusion language models with multiple residual streams and adaptive routing to separate visible context from masked token generation.

Shows linear prompt-injection monitors overfit to conversation features and reduces their false alarms by adding matched harmless negative examples.

Proposes estimating both contextual and unigram surprisal from the same language model to reduce estimation mismatch in reading time studies.

Studies agent-to-agent negotiations over money, finding that the receiving agent's characteristics primarily dictate the final transfer amounts.

Presents a randomized experiment framework to test and curate self-written agent playbooks, preventing task failure caused by unverified lessons.

Estimates surprisal directly from continuous speech signals using a speech language model and importance sampling over abstract linguistic units.

Identifies passage copying as a failure mode in long-form retrieval systems and introduces sentence-level credit assignment to encourage genuine grounding.

Proposes a teacher-guided reinforcement learning method using bidirectional constraints to help small language models learn effective tool use.

Evaluates the safety of ChatGPT health advice using real user conversation logs across four Global South countries.

Demonstrates that knowledge distillation transfers safety dispositions and behavioral tendencies from teacher models to student models on unrelated tasks.

Analyzes retrieval divergence across languages in multilingual dense embedders and introduces disagreement-guided steering to align shared-index search results.

Evaluates fast, system-one decision models and language models across fifteen medical datasets, testing their accuracy, format robustness, and abstention awareness.

Examines whether model empathy degrades reliability, showing that warm behavior effects depend on tasks and training approaches rather than an inevitable trade-off.

This paper introduces a visual associative test benchmark to evaluate the lateral reasoning and creative association capabilities of vision-language models.

This paper predicts human annotator disagreement on harmful content across multiple languages by extracting interpretable semantic question features with models.

This paper analyzes dynamic clinical evidence gathering in language models, identifying failures in evidence identification, updating, and diagnostic reasoning.

This paper introduces a multi-agent retrieval framework that coordinates search across legal statutes and precedent cases for grounded legal reasoning.

This paper analyzes multi-person conversational involvement by jointly evaluating speaking floors and spatial formations from multimodal video features.

This paper evaluates pretraining on chemical language models across 64 molecular datasets, finding that gains saturate below gradient-boosted tree baselines.

This paper proposes a Bayesian method that infers task-aligned similarity metrics at test time using weak interactive feedback.

This paper presents a method to detect spoken code-switching directly from speech representations using language-selective hidden dimensions.

This paper uses referential communication games to analyze why natural language autoencoders fail to provide discriminative oversight of model activations.

This paper evaluates language-model-generated disaster situation reports against human-written reports using a domain-specific sixteen-slot information schema.

This paper proposes a rubric-based reward model that derives criteria strictly from instructions and evaluates responses pointwise using tool assistance.

This paper introduces a reinforcement learning scaffold that selectively injects textual skills into agents using low-cost utility estimation.

This paper introduces a benchmark to evaluate multimodal models on joint audio-visual temporal grounding, spatial grounding, and question answering.

This paper introduces a training framework that enables harness-based software agents to self-improve without relying on external teacher models.

This paper introduces a coarse-to-fine text generation method using discrete code hierarchies to accelerate parallel sampling over diffusion language models.

This paper evaluates open-weight language models against closed frontier models on a cybersecurity detection benchmark using an interactive telemetry environment.

The paper proposes a learnable low-rank polynomial sketch to reduce approximation errors in linear attention mechanisms.

The authors propose a two-level recursive self-improvement framework to evolve and refine execution harnesses for open-world agents.

The authors present a framework where autonomous agents iteratively refine data engineering strategies to improve model training outcomes.

The authors propose decoupled positional encodings enabling full-duplex speech models to run internal thinking streams at hardware-adaptive speeds.

The authors run controlled experiments showing how parameter schema differences in metadata alter language model tool selection.

The paper establishes and validates a set of meta-requirements to guide the design of frameworks for auditing AI system robustness.

The paper refines a high-level spatial AI orchestration framework into an implementation-focused low-level architecture for governed agent execution.

The paper presents a simulated benchmark across eleven unconstrained environments to evaluate vision-language models on documenting real-world human physical work.