PyTorch compiler that accelerates training and inference. Get built-in optimizations for performance, memory, parallelism, and easily write your own.
☆1,449Mar 17, 2026Updated this week
Alternatives and similar repositories for lightning-thunder
Users that are interested in lightning-thunder are comparing it to the libraries listed below
Sorting:
- Speed up model training by fixing data loading.☆580Mar 2, 2026Updated 2 weeks ago
- A Fusion Code Generator for NVIDIA GPUs (commonly known as "nvFuser")☆383Updated this week
- PyTorch native quantization and sparsity for training and inference☆2,739Updated this week
- 20+ high-performance LLMs with recipes to pretrain, finetune and deploy at scale.☆13,228Mar 6, 2026Updated 2 weeks ago
- A minimal Python framework for building custom AI inference servers with full control over logic, batching, and scaling.☆3,814Mar 2, 2026Updated 2 weeks ago
- PyTorch native post-training library☆5,707Updated this week
- TensorDict is a pytorch dedicated tensor container.☆1,015Updated this week
- A PyTorch native platform for training generative AI models☆5,162Updated this week
- Simple and efficient pytorch-native transformer text generation in <1000 LOC of python.☆6,187Aug 22, 2025Updated 6 months ago
- Minimalistic large language model 3D-parallelism training☆2,617Feb 19, 2026Updated last month
- Freeing data processing from scripting madness by providing a set of platform-agnostic customizable pipeline processing blocks.☆2,956Updated this week
- Schedule-Free Optimization in PyTorch☆2,265May 21, 2025Updated 10 months ago
- Machine learning metrics for distributed, scalable PyTorch applications.☆2,419Updated this week
- Tile primitives for speedy kernels☆3,232Updated this week
- Accessible large language models via k-bit quantization for PyTorch.☆8,052Updated this week
- A library for accelerating Transformer models on NVIDIA GPUs, including using 8-bit and 4-bit floating point (FP8 and FP4) precision on H…☆3,231Updated this week
- Fast and memory-efficient exact attention☆22,832Updated this week
- Efficient Triton Kernels for LLM Training☆6,216Updated this week
- Development repository for the Triton language and compiler☆18,708Updated this week
- Pretrain, finetune ANY AI model of ANY size on 1 or 10,000+ GPUs with zero code changes.☆30,926Mar 10, 2026Updated last week
- Puzzles for learning Triton☆2,336Updated this week
- Tools for merging pretrained large language models.☆6,867Mar 15, 2026Updated last week
- Hackable and optimized Transformers building blocks, supporting a composable construction.☆10,373Updated this week
- Save, load, host, and share AI model checkpoints without slowing down training. Host on Lightning AI or your own cloud with enterprise-gr…☆42Mar 2, 2026Updated 2 weeks ago
- Machine Learning Engineering Open Book☆17,440Updated this week
- PyTorch extensions for high performance and large scale training.☆3,403Apr 26, 2025Updated 10 months ago
- Sparsity-aware deep learning inference runtime for CPUs☆3,163Jun 2, 2025Updated 9 months ago
- Foundation Architecture for (M)LLMs☆3,135Apr 11, 2024Updated last year
- Official repository of Evolutionary Optimization of Model Merging Recipes☆1,406Nov 29, 2024Updated last year
- AITemplate is a Python framework which renders neural network into high performance CUDA/HIP C++ code. Specialized for FP16 TensorCore (N…☆4,709Updated this week
- Flexible and powerful tensor operations for readable and reliable code (for pytorch, jax, TF and others)☆9,430Feb 20, 2026Updated last month
- A pytorch quantization backend for optimum☆1,032Nov 21, 2025Updated 4 months ago
- Implementation of the LLaMA language model based on nanoGPT. Supports flash attention, Int8 and GPTQ 4bit quantization, LoRA and LLaMA-Ad…☆6,082Jul 1, 2025Updated 8 months ago
- Placeholder for the opensource Grid AI components☆45Jun 6, 2022Updated 3 years ago
- SGLang is a high-performance serving framework for large language models and multimodal models.☆24,829Updated this week
- LLM training in simple, raw C/CUDA☆29,216Jun 26, 2025Updated 8 months ago
- Go ahead and axolotl questions☆11,460Updated this week
- Fast, flexible LLM inference☆6,713Updated this week
- Minimal, clean code for the Byte Pair Encoding (BPE) algorithm commonly used in LLM tokenization.☆10,375Jul 1, 2024Updated last year