ROCm / flash-attention
Fast and memory-efficient exact attention
☆160Updated this week
Alternatives and similar repositories for flash-attention:
Users that are interested in flash-attention are comparing it to the libraries listed below
- An efficient GPU support for LLM inference with x-bit quantization (e.g. FP6,FP5).☆238Updated 4 months ago
- Applied AI experiments and examples for PyTorch☆243Updated this week
- Development repository for the Triton language and compiler☆109Updated this week
- GPTQ inference Triton kernel☆297Updated last year
- Official repository for LightSeq: Sequence Level Parallelism for Distributed Training of Long Context Transformers☆207Updated 6 months ago
- Fast Matrix Multiplications for Lookup Table-Quantized LLMs☆231Updated 2 weeks ago
- Fast low-bit matmul kernels in Triton☆257Updated this week
- A high-throughput and memory-efficient inference and serving engine for LLMs☆68Updated this week
- BitBLAS is a library to support mixed-precision matrix multiplications, especially for quantized LLM deployment.☆541Updated 3 weeks ago
- Fast Hadamard transform in CUDA, with a PyTorch interface☆151Updated 9 months ago
- ☆179Updated 5 months ago
- This repository contains the experimental PyTorch native float8 training UX☆222Updated 7 months ago
- High-speed GEMV kernels, at most 2.7x speedup compared to pytorch baseline.☆100Updated 8 months ago
- ☆100Updated 6 months ago
- 8-bit CUDA functions for PyTorch☆44Updated 2 weeks ago
- ☆112Updated 2 months ago
- A collection of memory efficient attention operators implemented in the Triton language.☆250Updated 9 months ago
- Boosting 4-bit inference kernels with 2:4 Sparsity☆67Updated 6 months ago
- ☆73Updated 4 months ago
- PyTorch bindings for CUTLASS grouped GEMM.☆103Updated 2 months ago
- [NeurIPS 2024] KVQuant: Towards 10 Million Context Length LLM Inference with KV Cache Quantization☆335Updated 7 months ago
- ☆187Updated 8 months ago
- ☆87Updated 6 months ago
- PyTorch bindings for CUTLASS grouped GEMM.☆70Updated 4 months ago
- [ICLR2025] Breaking Throughput-Latency Trade-off for Long Sequences with Speculative Decoding☆109Updated 3 months ago
- The official implementation of the EMNLP 2023 paper LLM-FP4☆187Updated last year
- Flash-LLM: Enabling Cost-Effective and Highly-Efficient Large Generative Model Inference with Unstructured Sparsity☆201Updated last year
- [MLSys'24] Atom: Low-bit Quantization for Efficient and Accurate LLM Serving☆298Updated 8 months ago
- Collection of kernels written in Triton language☆110Updated 3 weeks ago
- Dynamic Memory Management for Serving LLMs without PagedAttention☆308Updated 3 weeks ago