shawntan / scattermoeLinks
Triton-based implementation of Sparse Mixture of Experts.
☆246Updated 3 weeks ago
Alternatives and similar repositories for scattermoe
Users that are interested in scattermoe are comparing it to the libraries listed below
Sorting:
- Official repository for DistFlashAttn: Distributed Memory-efficient Attention for Long-context LLMs Training☆216Updated last year
- ☆112Updated last year
- ☆145Updated 8 months ago
- PyTorch bindings for CUTLASS grouped GEMM.☆124Updated 4 months ago
- ☆149Updated 2 years ago
- ring-attention experiments☆154Updated last year
- 🚀 Efficiently (pre)training foundation models with native PyTorch features, including FSDP for training and SDPA implementation of Flash…☆270Updated 2 months ago
- Boosting 4-bit inference kernels with 2:4 Sparsity☆83Updated last year
- ☆130Updated 4 months ago
- This repository contains the experimental PyTorch native float8 training UX☆223Updated last year
- Collection of kernels written in Triton language☆157Updated 6 months ago
- GEAR: An Efficient KV Cache Compression Recipefor Near-Lossless Generative Inference of LLM☆169Updated last year
- Applied AI experiments and examples for PyTorch☆299Updated 2 months ago
- 🔥 A minimal training framework for scaling FLA models☆266Updated last month
- Fast and memory-efficient exact attention☆71Updated 7 months ago
- Flash-Muon: An Efficient Implementation of Muon Optimizer☆195Updated 4 months ago
- 🚀 Collection of components for development, training, tuning, and inference of foundation models leveraging PyTorch native components.☆215Updated last week
- Fast low-bit matmul kernels in Triton☆381Updated 3 weeks ago
- Explorations into some recent techniques surrounding speculative decoding☆288Updated 10 months ago
- Cold Compress is a hackable, lightweight, and open-source toolkit for creating and benchmarking cache compression methods built on top of…☆147Updated last year
- [NeurIPS 2024] KVQuant: Towards 10 Million Context Length LLM Inference with KV Cache Quantization☆389Updated last year
- Code for exploring Based models from "Simple linear attention language models balance the recall-throughput tradeoff"☆241Updated 4 months ago
- ☆121Updated last year
- Best practices for training DeepSeek, Mixtral, Qwen and other MoE models using Megatron Core.☆111Updated last week
- Cataloging released Triton kernels.☆263Updated last month
- extensible collectives library in triton☆89Updated 6 months ago
- Tritonbench is a collection of PyTorch custom operators with example inputs to measure their performance.☆264Updated this week
- [MLSys'24] Atom: Low-bit Quantization for Efficient and Accurate LLM Serving☆320Updated last year
- Triton implementation of FlashAttention2 that adds Custom Masks.☆141Updated last year
- [ICLR2025] Breaking Throughput-Latency Trade-off for Long Sequences with Speculative Decoding☆130Updated 10 months ago