nshepperd / flash_attn_jaxLinks
JAX bindings for Flash Attention v2
☆97Updated this week
Alternatives and similar repositories for flash_attn_jax
Users that are interested in flash_attn_jax are comparing it to the libraries listed below
Sorting:
- Accelerated First Order Parallel Associative Scan☆189Updated last year
- A library for unit scaling in PyTorch☆130Updated 3 months ago
- seqax = sequence modeling + JAX☆168Updated 3 months ago
- ☆149Updated 2 years ago
- Understand and test language model architectures on synthetic tasks.☆233Updated 3 weeks ago
- Experiment of using Tangent to autodiff triton☆80Updated last year
- supporting pytorch FSDP for optimizers☆83Updated 10 months ago
- Supporting code for the blog post on modular manifolds.☆86Updated 3 weeks ago
- Flash-Muon: An Efficient Implementation of Muon Optimizer☆195Updated 4 months ago
- PyTorch-native post-training at scale☆83Updated this week
- ☆121Updated last year
- ☆91Updated last year
- A MAD laboratory to improve AI architecture designs 🧪☆131Updated 10 months ago
- A simple library for scaling up JAX programs☆144Updated 11 months ago
- Code for exploring Based models from "Simple linear attention language models balance the recall-throughput tradeoff"☆241Updated 4 months ago
- Tree Attention: Topology-aware Decoding for Long-Context Attention on GPU clusters☆130Updated 10 months ago
- Fast and memory-efficient exact attention☆71Updated 7 months ago
- Minimal but scalable implementation of large language models in JAX☆35Updated last month
- The simplest, fastest repository for training/finetuning medium-sized GPTs.☆166Updated 3 months ago
- ☆251Updated 4 months ago
- Implementation of Flash Attention in Jax☆219Updated last year
- Triton-based implementation of Sparse Mixture of Experts.☆246Updated 3 weeks ago
- LoRA for arbitrary JAX models and functions☆141Updated last year
- A fusion of a linear layer and a cross entropy loss, written for pytorch in triton.☆69Updated last year
- A bunch of kernels that might make stuff slower 😉☆62Updated this week
- ☆41Updated last week
- ring-attention experiments☆154Updated last year
- Normalized Transformer (nGPT)☆192Updated 11 months ago
- CUDA and Triton implementations of Flash Attention with SoftmaxN.☆73Updated last year
- Minimal (400 LOC) implementation Maximum (multi-node, FSDP) GPT training☆132Updated last year