Popular repositories Loading

1. gptq gptq Public

   Code for the ICLR 2023 paper "GPTQ: Accurate Post-training Quantization of Generative Pretrained Transformers".

   Python 1.9k 150

2. sparsegpt sparsegpt Public

   Code for the ICML 2023 paper "SparseGPT: Massive Language Models Can Be Accurately Pruned in One-Shot".

   Python 694 88

3. marlin marlin Public

   FP16xINT4 LLM inference kernel that can achieve near-ideal ~4x speedups up to medium batchsizes of 16-32 tokens.

   Python 555 44

4. qmoe qmoe Public

   Code for the paper "QMoE: Practical Sub-1-Bit Compression of Trillion-Parameter Models".

   Python 258 22

5. PanzaMail PanzaMail Public

   Python 254 13

6. QUIK QUIK Public

   Repository for the QUIK project, enabling the use of 4bit kernels for generative inference

   C++ 167 12