Jie Zhou

S KILL G YM : I NTERNALIZING L ARGE -S CALE H UMAN S KILLS INTO LLM S FOR R EAL -W ORLD P ROBLEM S OLVING

WEFT: Scaling Tool-Use Post-Training for General-Purpose Agents

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, task, agent harness, and evaluator. Scaling environments in isolation, however, does not guarantee com...

SkillGym: Internalizing Human Skills into LLMs for Real-World Problem Solving

35B \texttt{SkillGym-Agent} reaches 51.47\% on skill-assisted SkillsBench, exceeding reported scores for Claude Sonnet 4.6, GPT-5.4 Mini, and DeepSeek V4 Pro, suggesting reusable procedural competence.

AutoMem: A Text-Gradient Recursive Self-Improvement Framework for Automated Memory Architectures Search

This work proposes \textsc{AutoMem}, a text-gradient recursive self-improvement framework for task-adaptive memory architecture search that consistently discovers task-adaptive memory architectures that outperform the strongest human-designed memory baselines.

AgentHPOBench: A Benchmark For Evaluating LLM Agents as Sequential Hyperparameter Optimizers

AgentHPOBench, a sequential benchmark comprising 30 executable machine learning tasks across seven research categories, shows that current agents exhibit measurable experimental optimization ability across domains, but still face clear limitations in sustained iterative refinement, complex log diagnosis, and consistent...