meta-pytorch / attention-gymLinks
Helpful tools and examples for working with flex-attention
β991Updated 2 weeks ago
Alternatives and similar repositories for attention-gym
Users that are interested in attention-gym are comparing it to the libraries listed below
Sorting:
- Implementation of π Ring Attention, from Liu et al. at Berkeley AI, in Pytorchβ539Updated 4 months ago
- A subset of PyTorch's neural network modules, written in Python using OpenAI's Triton.β576Updated last month
- π³ Efficient Triton implementations for "Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention"β870Updated 6 months ago
- Large Context Attentionβ740Updated 8 months ago
- Muon is an optimizer for hidden layers in neural networksβ1,735Updated 2 months ago
- Implementation of the sparse attention pattern proposed by the Deepseek team in their "Native Sparse Attention" paperβ753Updated last month
- β527Updated this week
- Ring attention implementation with flash attentionβ873Updated 2 weeks ago
- Pipeline Parallelism for PyTorchβ779Updated last year
- LLM KV cache compression made easyβ613Updated this week
- Scalable and Performant Data Loadingβ304Updated last week
- Annotated version of the Mamba paperβ490Updated last year
- [ICLR2025 Spotlightπ₯] Official Implementation of TokenFormer: Rethinking Transformer Scaling with Tokenized Model Parametersβ573Updated 7 months ago
- Code for Adam-mini: Use Fewer Learning Rates To Gain More https://arxiv.org/abs/2406.16793β437Updated 4 months ago
- A Quirky Assortment of CuTe Kernelsβ582Updated this week
- For optimization algorithm research and development.β538Updated last week
- β725Updated 2 weeks ago
- Tutel MoE: Optimized Mixture-of-Experts Library, Support GptOss/DeepSeek/Kimi-K2/Qwen3 using FP8/NVFP4/MXFP4β924Updated last week
- π Efficiently (pre)training foundation models with native PyTorch features, including FSDP for training and SDPA implementation of Flashβ¦β268Updated 2 months ago
- [ICLR 2025] Official PyTorch Implementation of Gated Delta Networks: Improving Mamba2 with Delta Ruleβ297Updated last week
- Implementation of Rotary Embeddings, from the Roformer paper, in Pytorchβ751Updated 2 months ago
- Fault tolerance for PyTorch (HSDP, LocalSGD, DiLoCo, Streaming DiLoCo)β404Updated last week
- π₯ A minimal training framework for scaling FLA modelsβ250Updated 2 weeks ago
- β302Updated 5 months ago
- Building blocks for foundation models.β556Updated last year
- π Efficient implementations of state-of-the-art linear attention modelsβ3,341Updated last week
- Load compute kernels from the Hubβ287Updated this week
- FlashFFTConv: Efficient Convolutions for Long Sequences with Tensor Coresβ330Updated 8 months ago
- A Distributed Attention Towards Linear Scalability for Ultra-Long Context, Heterogeneous Data Trainingβ515Updated this week
- Microsoft Automatic Mixed Precision Libraryβ620Updated 11 months ago