FlagOpen / FlagGems
FlagGems is an operator library for large language models implemented in Triton Language.
☆328Updated this week
Related projects ⓘ
Alternatives and complementary repositories for FlagGems
- A collection of memory efficient attention operators implemented in the Triton language.☆215Updated 5 months ago
- A Easy-to-understand TensorOp Matmul Tutorial☆287Updated last month
- flash attention tutorial written in python, triton, cuda, cutlass☆194Updated 4 months ago
- Dynamic Memory Management for Serving LLMs without PagedAttention☆222Updated this week
- A fast communication-overlapping library for tensor parallelism on GPUs.☆217Updated last week
- Disaggregated serving system for Large Language Models (LLMs).☆348Updated 2 months ago
- ☆136Updated this week
- ☆140Updated 6 months ago
- Several optimization methods of half-precision general matrix multiplication (HGEMM) using tensor core with WMMA API and MMA PTX instruct…☆296Updated 2 months ago
- Analyze the inference of Large Language Models (LLMs). Analyze aspects like computation, storage, transmission, and hardware roofline mod…☆310Updated last month
- ☆162Updated 3 months ago
- QServe: W4A8KV4 Quantization and System Co-design for Efficient LLM Serving☆433Updated 2 months ago
- ☆79Updated 2 months ago
- [EMNLP 2024 Industry Track] This is the official PyTorch implementation of "LLMC: Benchmarking Large Language Model Quantization with a V…☆315Updated this week
- optimized BERT transformer inference on NVIDIA GPU. https://arxiv.org/abs/2210.03052☆457Updated 7 months ago
- A low-latency & high-throughput serving engine for LLMs☆231Updated last month
- ☆123Updated this week
- A simple high performance CUDA GEMM implementation.☆334Updated 10 months ago
- USP: Unified (a.k.a. Hybrid, 2D) Sequence Parallel Attention for Long Context Transformers Model Training and Inference☆352Updated last week
- BitBLAS is a library to support mixed-precision matrix multiplications, especially for quantized LLM deployment.☆407Updated this week
- Development repository for the Triton-Linalg conversion☆148Updated 3 weeks ago
- ☆189Updated last month
- An efficient GPU support for LLM inference with x-bit quantization (e.g. FP6,FP5).☆196Updated last week
- Performance of the C++ interface of flash attention and flash attention v2 in large language model (LLM) inference scenarios.☆26Updated 2 months ago
- ☆56Updated this week
- learning how CUDA works☆162Updated 2 months ago
- FlagScale is a large model toolkit based on open-sourced projects.☆167Updated this week
- Shared Middle-Layer for Triton Compilation☆185Updated this week
- ☆283Updated 7 months ago
- Zero Bubble Pipeline Parallelism☆279Updated this week