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 utilization in both training and inference.
☆2,311Updated this week
Alternatives and similar repositories for TransformerEngine:
Users that are interested in TransformerEngine are comparing it to the libraries listed below
- FlashInfer: Kernel Library for LLM Serving☆2,532Updated this week
- [MLSys 2024 Best Paper Award] AWQ: Activation-aware Weight Quantization for LLM Compression and Acceleration☆2,895Updated last week
- A unified library of state-of-the-art model optimization techniques such as quantization, pruning, distillation, speculative decoding, et…☆832Updated 2 weeks ago
- [ICML 2023] SmoothQuant: Accurate and Efficient Post-Training Quantization for Large Language Models☆1,373Updated 8 months ago
- Transformer related optimization, including BERT, GPT☆6,100Updated last year
- PyTorch native quantization and sparsity for training and inference☆1,927Updated this week
- Pipeline Parallelism for PyTorch☆761Updated 7 months ago
- Automatically Discovering Fast Parallelization Strategies for Distributed Deep Neural Network Training☆1,779Updated last week
- Ongoing research training transformer language models at scale, including: BERT & GPT-2☆2,034Updated last week
- Microsoft Automatic Mixed Precision Library☆587Updated 6 months ago
- [ICML 2024] Break the Sequential Dependency of LLM Inference Using Lookahead Decoding☆1,221Updated 3 weeks ago
- MII makes low-latency and high-throughput inference possible, powered by DeepSpeed.☆1,998Updated last week
- FP16xINT4 LLM inference kernel that can achieve near-ideal ~4x speedups up to medium batchsizes of 16-32 tokens.☆783Updated 6 months ago
- Code for the ICLR 2023 paper "GPTQ: Accurate Post-training Quantization of Generative Pretrained Transformers".☆2,067Updated last year
- A PyTorch Native LLM Training Framework☆763Updated 3 months ago
- Tile primitives for speedy kernels☆2,208Updated this week
- SOTA low-bit LLM quantization (INT8/FP8/INT4/FP4/NF4) & sparsity; leading model compression techniques on TensorFlow, PyTorch, and ONNX R…☆2,365Updated this week
- The Triton TensorRT-LLM Backend☆816Updated this week
- A throughput-oriented high-performance serving framework for LLMs☆785Updated 6 months ago
- NCCL Tests☆1,047Updated 2 weeks ago
- A Python-level JIT compiler designed to make unmodified PyTorch programs faster.☆1,038Updated 11 months ago
- PyTorch extensions for high performance and large scale training.☆3,285Updated 2 months ago
- Kernl lets you run PyTorch transformer models several times faster on GPU with a single line of code, and is designed to be easily hackab…☆1,560Updated last year
- Official Implementation of EAGLE-1 (ICML'24), EAGLE-2 (EMNLP'24), and EAGLE-3.☆1,113Updated last week
- Domain-specific language designed to streamline the development of high-performance GPU/CPU/Accelerators kernels☆858Updated this week
- AutoAWQ implements the AWQ algorithm for 4-bit quantization with a 2x speedup during inference. Documentation:☆2,048Updated 3 weeks ago
- Minimalistic large language model 3D-parallelism training☆1,737Updated this week
- Ring attention implementation with flash attention☆721Updated last month
- BitBLAS is a library to support mixed-precision matrix multiplications, especially for quantized LLM deployment.☆566Updated last month
- A pytorch quantization backend for optimum☆910Updated 3 weeks ago