by zengxiao-he · LLM 插件 · ★ 579
Tessera A small, from-scratch LLM stack built around one goal: distill a large teacher into a small student, then serve that student efficiently. Keeping that goal end-to-end means touching most of the pieces that matter in practice — custom GPU kernels, sharded training, an inference engine, quantization, and a serving front end — without any of it being a toy. It runs and is unit-tested on a laptop (CPU or Apple MPS).
| Stars | 579 |
| Forks | 9 |
| Language | Python |
| Category | LLM 插件 |
| Quality Score | 62.4904431562059/100 |
| Last Updated | 2026-06-05 |
| Created | 2026-06-05 |
| Platforms | python |
| Est. Tokens | ~7k |
These tools work well together with tessera for enhanced workflows:
Explore other popular llm 插件 tools:
tessera is From teacher to tiles — a from-scratch LLM distillation & serving engine: custom Triton/CUDA kernels, FSDP distillation, paged-KV continuous batching, speculative decoding, a Rust gateway, a JAX oracl. It is categorized as a LLM 插件 with 579 GitHub stars.
tessera is primarily written in Python. It covers topics such as cuda, flash-attention, fsdp.
You can find installation instructions and usage details in the tessera GitHub repository at github.com/zengxiao-he/tessera. The project has 579 stars and 9 forks, indicating an active community.