Zhiheng Xi

A Token-Level Analysis of Sampled-Token Reverse-KL On-Policy Distillation

On-policy distillation (OPD) supervises a student on its own trajectories with token-level signals from a frozen teacher, yet how a sampled loss allocates updates across tokens remains poorly understood. We analyze the gradient of the per-token K2 estimator of reverse KL with respect to the student logits. The $\ell_1$...

Atria Dawn: The Dawn of Agentic Superintelligence

As AI agents become participants in the development of their successors, they reshape both the production of intelligence and the role of human researchers. We introduce Atria Dawn Preview, a foundation agentic language model designed for scientific research and engineering workflows, with the goal of expanding the fro...

MathCritique: Enhancing LLM Reasoning via Critique Models with Test-Time and Training-Time Supervision

Prefix-Adaptive Block Diffusion for Efficient Document Recognition

The Prefix-Adaptive Block Diffusion Model (PA-BDM) is proposed, which replaces intra-block bidirectional denoising with causal denoising from prefix to suffix and treats the block size as a maximum candidate range rather than a fixed commitment unit.

JFTA-Bench: Evaluate LLM's Ability of Tracking and Analyzing Malfunctions Using Fault Trees

A novel textual representation of fault trees is proposed, and a benchmark for multi-turn dialogue systems that emphasizes robust interaction in complex environments is constructed, evaluating a model's ability to assist in malfunction localization.

CAFE: Self-Improving Search Agents Need Co-Evolving Feedback

CAFE (Coupled Agent--Feedback Evolution), a framework in which a shared-parameter model alternates between search-agent and critic roles, is introduced, suggesting that a self-improving search agent needs feedback that co-evolves with the policy it guides.

Counteracting the Matthew Effect in Self-Improvement of LVLMs through Head-Tail Re-balancing

To mitigate a critical imbalance during the exploration-and-learning process, this work approaches head-tail re-balance during the exploration-and-learning process from two perspectives: distribution-reshaping and trajectory-resampling.

AgentGym2: Benchmarking Large Language Model Agents in De-Idealized Real-World Environments

AgentGym2 is presented, a new evaluation framework with task instances grounded in real-world end-to-end working demands that measures agents'ability to execute end-to-end procedures, discover tools via exploration, compose tools for unseen tasks, and remain robust to noisy and underspecified information.