Popular repositories Loading

1. TransformerEngine TransformerEngine Public

   Forked from NVIDIA/TransformerEngine

   A library for accelerating Transformer models on NVIDIA GPUs, including using 8-bit floating point (FP8) precision on Hopper and Ada GPUs, to provide better performance with lower memory utilizatio…

   Python 1

2. Megatron-LM Megatron-LM Public

   Forked from NVIDIA/Megatron-LM

   Ongoing research training transformer models at scale

   Python

3. Model-Optimizer Model-Optimizer Public

   Forked from NVIDIA/Model-Optimizer

   A unified library of SOTA model optimization techniques like quantization, pruning, distillation, speculative decoding, etc. It compresses deep learning models for downstream deployment frameworks …

   Python