Zhongqi Sun

A unified library of SOTA model optimization techniques like quantization, pruning, distillation, speculative decoding, etc. It compresses deep learning models for downstream deployment frameworks like TensorRT-LLM, TensorRT, vLLM, etc. to optimize inference speed.

Ultra-high-performance, secure, all-in-one acceleration engine for developer resources

《Effective Modern C++》- 学习记录

Deep Learning System core principles introduction.