Ph.D. student @ HKUST(GZ) · Research Intern @ Mind Lab
Efficient and reliable LLMs: inference, long context, KV cache, retrieval, and agentic workflows.

40+ stars personal public non-fork repos

8.4k+ / 1.1k+ contributed projects: LMFlow / kvpress

benchmark → method → artifact how I like research to ship

Current Focus

Inference efficiency KV-cache compression, token-efficient reasoning, energy-to-token evaluation, serving bottlenecks.	Long-context evaluation Generation-focused benchmarks, dense reasoning integrity, multi-turn coherence.
Agent systems Tool use, post-training, harness design, local-first agent workflow infrastructure.	Research infrastructure Reproducible artifacts, project pages, scholar tracking, figure and report tooling.

Selected Work

LMFlow Contributed to an extensible toolkit for fine-tuning and inference of large foundation models.	LongGenBench Long-context generation benchmark for coherent, context-aware long-form responses.
QuantArena Policy-conditioned live-market evaluation for LLM trading agents. Benchmark the policy, not just the model.	agent-hub Local-first agent task hub with SQLite queueing, dependency-aware dispatch, templates, and dashboards.
tinker2openai-tool Adapters between XML-like tool calls and OpenAI-style structured tool-call histories.	energy-to-token Project page for evaluating LLM inference as energy-to-token production.

Research Map

long-context generation ──┬── LongGenBench
                          ├── semantic integrity under KV compression
                          └── multi-turn coherence / FlowKV

agent capability eval ────┬── QuantArena
                          ├── tool-use adapters
                          └── local-first agent workflow runtime

efficient inference ──────┬── ChunkKV / KV compression
                          ├── token-efficient reasoning
                          └── energy-to-token production

Stack

repositories · publications · citations