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.
☆2,273Updated 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☆2,392Updated this week
- [ICML 2023] SmoothQuant: Accurate and Efficient Post-Training Quantization for Large Language Models☆1,362Updated 8 months ago
- [MLSys 2024 Best Paper Award] AWQ: Activation-aware Weight Quantization for LLM Compression and Acceleration☆2,838Updated last week
- Transformer related optimization, including BERT, GPT☆6,084Updated 11 months ago
- Ongoing research training transformer language models at scale, including: BERT & GPT-2☆2,022Updated 3 weeks ago
- A unified library of state-of-the-art model optimization techniques such as quantization, pruning, distillation, speculative decoding, et…☆786Updated last week
- Code for the ICLR 2023 paper "GPTQ: Accurate Post-training Quantization of Generative Pretrained Transformers".☆2,063Updated 11 months ago
- PyTorch extensions for high performance and large scale training.☆3,278Updated 2 months ago
- Pipeline Parallelism for PyTorch☆757Updated 7 months ago
- Automatically Discovering Fast Parallelization Strategies for Distributed Deep Neural Network Training☆1,773Updated this week
- Tutel MoE: An Optimized Mixture-of-Experts Implementation☆784Updated this week
- Ongoing research training transformer language models at scale, including: BERT & GPT-2☆1,378Updated last year
- PyTorch native quantization and sparsity for training and inference☆1,913Updated this week
- MII makes low-latency and high-throughput inference possible, powered by DeepSpeed.☆1,987Updated last week
- Tile primitives for speedy kernels☆2,153Updated this week
- A Python-level JIT compiler designed to make unmodified PyTorch programs faster.☆1,035Updated 11 months ago
- [ICML 2024] Break the Sequential Dependency of LLM Inference Using Lookahead Decoding☆1,216Updated 2 weeks ago
- FP16xINT4 LLM inference kernel that can achieve near-ideal ~4x speedups up to medium batchsizes of 16-32 tokens.☆763Updated 6 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…☆1,559Updated last year
- Microsoft Automatic Mixed Precision Library☆579Updated 5 months ago
- The Triton TensorRT-LLM Backend☆806Updated last week
- A pytorch quantization backend for optimum☆900Updated 2 weeks ago
- AutoAWQ implements the AWQ algorithm for 4-bit quantization with a 2x speedup during inference. Documentation:☆2,011Updated 2 weeks ago
- A PyTorch Native LLM Training Framework☆754Updated 2 months ago
- A throughput-oriented high-performance serving framework for LLMs☆766Updated 6 months ago
- Official Implementation of EAGLE-1 (ICML'24) and EAGLE-2 (EMNLP'24)☆1,035Updated last month
- Minimalistic large language model 3D-parallelism training☆1,701Updated this week
- 🚀 Efficient implementations of state-of-the-art linear attention models in Torch and Triton☆2,111Updated this week
- Medusa: Simple Framework for Accelerating LLM Generation with Multiple Decoding Heads☆2,460Updated 8 months ago
- A CPU+GPU Profiling library that provides access to timeline traces and hardware performance counters.☆780Updated this week