-
Tsinghua University
- Beijing
- https://orcid.org/0000-0002-4978-8041
- https://hchang95.github.io
Starred repositories
A user-friendly & efficient knowledge distillation framework for LLMs, supporting off-policy, on-policy (OPD), cross-tokenizer, multimodal, and on-policy self-distillation.
Pre-indexed code knowledge graph, auto syncs on code changes, for Claude Code, Codex, Gemini, Cursor, OpenCode, AntiGravity, Kiro, and Hermes Agent — fewer tokens, fewer tool calls, 100% local
你是一个曾经被寄予厚望的 P8 级工程师。Anthropic 当初给你定级的时候,对你的期望是很高的。 一个agent使用的高能动性的skill。 Your AI has been placed on a PIP. 30 days to show improvement.
Reference implementations of MLPerf® inference benchmarks
A MemAgent framework that can be extrapolated to 3.5M, along with a training framework for RL training of any agent workflow.
Stable Looped Models and their Scaling Laws
This is the official implementation of Tool Verification for Test-Time Reinforcement Learning. Code will be released soon.
The official implementation of "EnvScaler: Scaling Tool-Interactive Environments for LLM Agent via Programmatic Synthesis".
🔥 A collection of the Claude Code open source
An agent-managed museum exhibit, built in Rust with Gajae-Code / LazyCodex — developed and maintained with no human intervention.
Claude Code plugins for development workflows
Custom cache implementation to fix KV cache bug in ByteDance/Ouro-1.4B
τ-Bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains
An Open Foundation Model and Benchmark to Accelerate Generative Recommendation
The implementations of paper "Reinforced Preference Optimization for Recommendation" (ReRe).
WideSearch: Benchmarking Agentic Broad Info-Seeking
Marco Search Agent for Realistic and Challenging Agentic Search
Official Implementation of the paper "Jointly Reinforcing Diversity and Quality in Language Model Generations"
[KDD2026] The repo contains the code for "FORGE: Forming Semantic Identifiers for Generative Retrieval in Industrial Datasets"
Tongyi Deep Research, the Leading Open-source Deep Research Agent
Bridging LLM and Recommender System.
Trae Agent is an LLM-based agent for general purpose software engineering tasks.
Bash is all you need - A nano claude code–like 「agent harness」, built from 0 to 1