저장소
zengxiao-he/tessera
From teacher to tiles — a from-scratch LLM distillation & serving engine: custom Triton/CUDA kernels, FSDP distillati…
- Stars
- ★ 434
- Forks
- 5
- Issues
- 0
- Updated
- 6월 5일
- Language
- Python
- License
- NOASSERTION
#cuda#flash-attention#fsdp#inference-engine#jax#knowledge-distillation#kv-cache#llm