AI research papers.

Long-context reasoning in large language models incurs substantial computation and memory costs.

Recurrent reasoning models have attracted growing attention for scaling test-time computation, typically by iteratively refining latent states with shared parameters.

The strong generalization performance of on-policy post-training paradigms has motivated studies of their parameter update behaviors.

Embodied agents now take on ever longer tasks.

Manipulation behaviors vary widely across objects and scenes, but they share a small set of reusable skills, and planning with these skills helps embodied agents generalize to new tasks.

Agents increasingly build on code written by other agents, and they reimplement rather than reuse, growing the codebases later agents must work in.

Group Relative Policy Optimization (GRPO) is widely used to train reasoning language models, where it computes advantages by centering and normalizing rewards across rollouts of the same prompt.

Recent efforts to scale tool-use post-training have largely centered on the synthesis of executable environments, which constitute only one component of a broader agentic interaction system comprising the environment, ta…

Reproducibility is essential for scientific research, yet prior work shows that LLM outputs vary with hardware and batching.

Manipulating next-token probabilities during generation can bypass the safety alignment of large language models.

System One models such as Jev offer an efficient alternative to generative language models for tasks that require decisions rather than open-ended responses.

Few-step streaming audio--video generation requires both causal modeling and step distillation, yet standard training recipes face two context-related challenges.

Natural language autoencoders translate a language model's internal activations into readable explanations.

Distributional Diffusion Models (DDMs) replace the standard mean-prediction denoiser with a distributional denoiser trained via a scoring rule objective, learning a stochastic approximation to p(x_1 mid x_t) rather than…

Multi-reward guided reinforcement learning (i.e., RL) offers a promising way to improve joint audio-video diffusion models along several complementary objectives, including modality-specific quality, cross-modal semantic…

Answering questions and completing tasks over large document collections often requires connecting evidence spread across multiple documents, such as a project's approval recorded in one, its requirements in another, and…

An agent that uses tools typically responds to what the user explicitly asks, yet completing the task may require information the user never requested.

Proactive LLM agents can turn idle compute into useful support before users ask.

Vision-language-action (VLA) policies often fail when a robot's executed motion deviates from their commanded action.

Large language model (LLM) routing aims to assign each query to the most suitable model from a heterogeneous candidate pool, improving the quality--efficiency trade-off of LLM inference.

On-policy distillation (OPD) trains students using teacher feedback on their own sampled responses, yet how prompt choice shapes transfer across teacher-student pairs remains poorly understood.

Large language model (LLM) routing reduces serving cost by assigning each query to an appropriate model while preserving response quality.

Masked diffusion models (MDMs) generate sequences by progressively unmasking several tokens per denoising step, but their reverse process is typically factorized over positions, limiting sample quality in the few-step re…

Skills equip LLM agents with professional knowledge and guidance to complete long-horizon and complex tasks.

To serve as real-world personal assistants, streaming video models need persistent memory that retains past experiences for later use.

Audio-visual large language models (AVLLMs) have made remarkable progress in multimodal understanding and reasoning through interactions among visual, auditory, and linguistic information.

Rendering accurate Chinese text remains challenging for text-to-image models.

Autonomous agents have made rapid progress in general-purpose computer use, but reliable automation of professional industrial engineering remains out of reach, as engineering workflows demand reasoning over geometric an…

Active test-time adaptation (ATTA) improves robustness under distribution shift by updating a deployed model during inference while selectively querying supervision.

Optical Character Recognition (OCR) is evolving from plain-text transcription toward general visual intelligence, requiring models to recognize, localize, and reason over textual information in complex visual environment…

We introduce Context Language Models (CLMs), language models that natively manage their own context.

Transformer language models build predictions through successive residual updates, but how their representations become specific to an eventual outcome remains unclear.

Vision-language models (VLMs) are increasingly used in place of human annotators, making it important that substitutability tests reflect the model rather than incidental evaluation conditions.

How many tokens from its context does a language model actually use, and what determines that number?

On-policy self-distillation (OPSD) trains a student to match a privileged teacher distribution along its own sampled trajectory.

Agents can learn from past executions, but enabling different agents to reuse and build on one another's experience remains challenging.

Autoregressive (AR) video diffusion enables low-latency, streamable video generation, but prediction errors often accumulate over long rollouts.

Video understanding agents acquire evidence through an executable harness that controls what they observe and how they use those observations.

Tabular machine learning typically relies on per-dataset workflows, fitting tree ensembles or running AutoML searches from scratch for every task.

High-resolution video generation is expensive, as its cost grows rapidly with the number of spatiotemporal tokens.

Tabular foundation models achieve strong zero-shot accuracy on structured data by pretraining on synthetic tables, but they ignore the column names, task descriptions, and auxiliary files that carry dataset semantics.

Computer use agents (CUAs) have demonstrated strong capabilities in completing digital tasks.

An agent skill is a reusable, actionable natural-language artifact that guides an agent to perform a task effectively under a given harness.

Improving robot policies on new tasks without collecting additional expert demonstrations remains a central challenge in robot learning.

Real-world robot learning is constrained by the cost of collecting experience and evaluating candidate behaviors.

Vision-language-action (VLA) models have advanced robotic manipulation, but their zero-shot generalization in new tasks and environments remains limited, and their reliance on specialized training keeps them from benefit…

Policies with similar mean returns can differ sharply in rare failures, yet estimating lower-tail conditional value-at-risk (CVaR) accurately can require many costly rollouts.

Power-sharpened sampling is an inference-time alternative to reinforcement-learning (RL) post-training for enhancing reasoning in large language models (LLMs).

Long-form video understanding requires multimodal agents to iteratively gather evidence over many reasoning steps.

KV cache memory and bandwidth costs grow with context length and batch size, which limits efficient long-context inference.