All papers

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

The paper presents a framework using vision-language models to generate reusable visual planning code from a single practice scene.

The paper analyzes the minimum context size required for transformer models to perform in-context reinforcement learning without parameter updates.

This paper studies cross-entropy surprisal metrics, showing that predictability rankings of texts remain consistent across multiple small language models.

Proposes a multimodal agent providing grounded prompts and video clips to guide video editing models in professional sports motions.

The paper examines communicating calibrated action probabilities rather than direct actions to improve multi-agent collaboration under partial information.

The paper empirically analyzes echo state network variants on synthetic sequence-to-sequence pattern tasks.

The paper studies how repeated input measurements can degrade out-of-sample prediction quality in tabular foundation models.

This paper introduces an evaluation index to measure whether machine translation of emergency communication preserves critical safety actionability.

Proposes an advertising image editing method that identifies and selectively refines visual attributes driving click-through rates.

Introduces an auditing framework that checks counterfactual execution traces to ensure language model routing uses only legitimate information.

The paper develops an analytical framework to interpret and reverse-engineer incentives and cooperative behaviors in multi-agent artificial intelligence systems.

This paper presents a memory-efficient subspace optimization framework using residual correction to stabilize language model training.

The paper examines conceptual parallels between measurement science and machine learning uncertainty evaluation to propose a more unified assessment framework.

Proposes spatial-semantic token compression and layer routing to accelerate 3D vision-language model inference across multi-view inputs.

The paper introduces a framework to recalibrate, audit, and safely deploy confidence scores for language model outputs.

This paper introduces MolLean, a verifiable language and tool interface that grounds chemical reasoning in formal state transitions and explicit checks.

This paper introduces an efficient encoder-only protein language model incorporating modern architectural updates to lower training costs.

This paper compares observational feature discovery scores in sparse autoencoders against actual causal intervention effects on model outputs.

Distills geometric privilege into vision-language models during training to improve spatial reasoning without inference overhead.

Analyzes how continual learning methods affect demographic fairness over time across eleven deepfake generation techniques.

This paper proposes Claim Cards to make agent interpretations explicit when verifying ambiguous natural-language scientific claims.

The paper details the development, training data criteria, and performance of the open Japanese sovereign models, LLM-jp-4.

The paper evaluates how commercial language models value human time by measuring recommended delay compensation amounts across 193 different countries.

Introduces Dynamic Vector Decoding to efficiently map 2D and 3D perception coordinates into compact discrete tokens for multimodal LLMs.

The paper demonstrates vulnerabilities in unilateral weight obfuscation schemes used for on-device secure language model inference.

The paper investigates how team composition and organizational interdependence affect performance in self-evolving multi-agent systems.

The paper introduces a distillation framework that triggers teacher takeovers when student policies diverge during long-horizon reasoning.

The paper introduces rate-resolved text evaluation to turn masking rates in diffusion language models into diagnostic scores for text assessment.

Proposes Syn-Omni, a framework combining modality-expert LoRA paths and progressive routing to balance shared and modality-specific omnimodal embeddings.

The paper proposes a selective safety framework for vision-language models using counterfactual pairs and representation density estimation.

The paper formulates task routing across language model providers as an auction mechanism to encourage truthful cost reporting while minimizing context switching.

This paper introduces FOLD, a training-free speculative decoding approach treating distant tokens as position-free far memory for long-context generation.

The paper introduces a Persian vision-language-action dataset and benchmark to evaluate robotic instruction grounding under deliberate ambiguity and infeasibility.

Presents VibeEdit, an image editor conditioned on spatial canvas annotations to perform targeted object additions, modifications, or removals.

The paper proposes a method combining offline execution-verified prompt updates and test-time refinement to improve program generation in language models.

This paper introduces TreeSpark, a framework training semi-autoregressive drafters to generate candidate trees aligned with speculative verification.

The paper investigates how varying the detail and format of feedback affects machine translation refinement in large language models.

The paper introduces an auditing framework that identifies failure steps in multi-agent systems by evaluating how information is utilized across handoffs.

Presents a system combining representation learning and music alignment to generate editable aesthetic stage lighting cues for live performances.

The paper introduces a unified routing framework that learns both execution mode selection and model assignment across diverse tasks.

This paper presents RM-OPD, a framework that trains domain-specialized reward models via on-policy distillation into a generalist model.

The authors analyze how different optimizers leave distinct implicit bias fingerprints that influence training data reconstruction from model checkpoints.

Analyzes the emergence and behavior of attention sinks in text-to-image diffusion and autoregressive Transformers.

Tests language models on graph-based information retrieval and navigation using the Wikipedia Game across multiple languages and citation datasets.

The paper introduces a localized Japanese benchmark for evaluating tool-using conversational agents in customer-support scenarios.

The paper tests a heuristic to set the number of transformer attention heads using the intrinsic dimension of token representations.

This paper investigates attention heads responsible for contextual entrainment and evaluates whether their ablation transfers across different semantic relations.

Presents a two-stage visual-textual alignment and retrieval-augmented framework for generating structured radiology reports from medical scans.

Introduces SqueezeBench, a saturation-resilient benchmark for formal mathematical reasoning where new accepted proofs dynamically raise the target bounds.

The authors propose a learned steering agent that monitors execution context and intervenes to prevent goal drift in tool-using language model systems.

The paper introduces a speculative decoding system combining self-refinement and pipelined drafting to reduce wall-clock generation time.

This paper presents a post-hoc method that refines fine-tuned models by selectively reverting parameter updates along principal directions of base weights.

Builds an automated real-to-sim-to-real framework using 3D Gaussian Splatting to adapt robot manipulation policies to specific rooms.

Introduces MAGMA-Bench to evaluate long-horizon embodied agent reasoning during dynamic task execution failures and multi-robot coordination.

The paper builds state graphs across multiple agent rollouts to evaluate behavioral consistency and predict execution failures.

The paper presents a four-stage training pipeline combining preference optimization and distillation for instruction-following machine translation.

The authors propose a probe-based method to measure and calibrate systematic scoring biases in black-box LLM evaluation judges.

The report presents an omni-modal model family trained by scaling reinforcement learning throughput, environments, and agentic reward compute.

Presents a benchmark to evaluate language models on generating mathematical figures and diagram-assisted reasoning across frontier problems.