fzyzcjy / torch_utilsLinks
Utility scripts for PyTorch (e.g. Memory profiler that understands more low-level allocations such as NCCL)
☆44Updated 2 weeks ago
Alternatives and similar repositories for torch_utils
Users that are interested in torch_utils are comparing it to the libraries listed below
Sorting:
- PyTorch bindings for CUTLASS grouped GEMM.☆109Updated 2 months ago
- ☆61Updated 3 months ago
- ☆97Updated 11 months ago
- Odysseus: Playground of LLM Sequence Parallelism☆76Updated last year
- A simple calculation for LLM MFU.☆43Updated 5 months ago
- ☆50Updated 3 months ago
- PyTorch bindings for CUTLASS grouped GEMM.☆135Updated last month
- Bridge Megatron-Core to Hugging Face/Reinforcement Learning☆93Updated last week
- Estimate MFU for DeepSeekV3☆24Updated 7 months ago
- ☆92Updated 4 months ago
- DeeperGEMM: crazy optimized version☆71Updated 3 months ago
- TritonBench: Benchmarking Large Language Model Capabilities for Generating Triton Operators☆74Updated 2 months ago
- ☆78Updated 4 months ago
- Implement Flash Attention using Cute.☆94Updated 8 months ago
- Allow torch tensor memory to be released and resumed later☆109Updated last week
- Official repository for DistFlashAttn: Distributed Memory-efficient Attention for Long-context LLMs Training☆214Updated last year
- Best practices for training DeepSeek, Mixtral, Qwen and other MoE models using Megatron Core.☆56Updated this week
- 16-fold memory access reduction with nearly no loss☆104Updated 4 months ago
- ☆43Updated last year
- A Suite for Parallel Inference of Diffusion Transformers (DiTs) on multi-GPU Clusters☆47Updated last year
- nnScaler: Compiling DNN models for Parallel Training☆115Updated this week
- Quantized Attention on GPU☆44Updated 9 months ago
- Summary of system papers/frameworks/codes/tools on training or serving large model☆57Updated last year
- A lightweight design for computation-communication overlap.☆155Updated last week
- ☆77Updated 3 months ago
- ☆145Updated 5 months ago
- ☆91Updated this week
- ☆86Updated 9 months ago
- Flash-LLM: Enabling Cost-Effective and Highly-Efficient Large Generative Model Inference with Unstructured Sparsity☆217Updated last year
- Framework to reduce autotune overhead to zero for well known deployments.☆80Updated last week