google / paxml
Pax is a Jax-based machine learning framework for training large scale models. Pax allows for advanced and fully configurable experimentation and parallelization, and has demonstrated industry leading model flop utilization rates.
☆457Updated last week
Related projects ⓘ
Alternatives and complementary repositories for paxml
- ☆178Updated last week
- JAX-Toolbox☆245Updated this week
- ☆122Updated this week
- jax-triton contains integrations between JAX and OpenAI Triton☆343Updated 3 weeks ago
- ☆267Updated this week
- ☆224Updated 4 months ago
- Orbax provides common checkpointing and persistence utilities for JAX users☆303Updated this week
- ☆334Updated 7 months ago
- Legible, Scalable, Reproducible Foundation Models with Named Tensors and Jax☆516Updated this week
- Pipeline Parallelism for PyTorch☆726Updated 2 months ago
- ☆303Updated this week
- JetStream is a throughput and memory optimized engine for LLM inference on XLA devices, starting with TPUs (and GPUs in future -- PRs wel…☆232Updated this week
- seqax = sequence modeling + JAX☆133Updated 4 months ago
- JAX implementation of the Llama 2 model☆210Updated 9 months ago
- A library to analyze PyTorch traces.☆307Updated this week
- Implementation of Flash Attention in Jax☆196Updated 8 months ago
- CLU lets you write beautiful training loops in JAX.☆321Updated this week
- MLCommons Algorithmic Efficiency is a benchmark and competition measuring neural network training speedups due to algorithmic improvement…☆333Updated 3 weeks ago
- A subset of PyTorch's neural network modules, written in Python using OpenAI's Triton.☆483Updated 3 weeks ago
- Inference code for LLaMA models in JAX☆113Updated 6 months ago
- ☆197Updated 4 months ago
- Train very large language models in Jax.☆195Updated last year
- This repository contains the experimental PyTorch native float8 training UX☆211Updated 3 months ago
- JAX Synergistic Memory Inspector☆164Updated 4 months ago
- Task-based datasets, preprocessing, and evaluation for sequence models.☆561Updated this week
- Implementation of a Transformer, but completely in Triton☆249Updated 2 years ago
- TorchX is a universal job launcher for PyTorch applications. TorchX is designed to have fast iteration time for training/research and sup…☆332Updated this week
- ☆328Updated this week
- A performant, memory-efficient checkpointing library for PyTorch applications, designed with large, complex distributed workloads in mind…☆146Updated 2 weeks ago
- A Python-level JIT compiler designed to make unmodified PyTorch programs faster.☆1,011Updated 7 months ago