A set of Python scripts that makes your experience on TPU better

TPU pod commander is a package for managing and launching jobs on Google Cloud TPU pods.

Minimal but scalable implementation of large language models in JAX

Read Google Cloud Storage, Azure Blobs, and local paths with the same interface

Turn jitted jax functions back into python source code

[ICLR 2024] SWE-Bench: Can Language Models Resolve Real-world Github Issues?

lightweight, standalone C++ inference engine for Google's Gemma models.

Code for Paper: Autonomous Evaluation and Refinement of Digital Agents

LLM training in simple, raw C/CUDA

Schedule-Free Optimization in PyTorch

Reaching LLaMA2 Performance with 0.1M Dollars

SWE-agent takes a GitHub issue and tries to automatically fix it, using GPT-4, or your LM of choice. It solves 12.47% of bugs in the SWE-bench evaluation set and takes just 1 minute to run.

HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal

Legible, Scalable, Reproducible Foundation Models with Named Tensors and Jax

Language models scale reliably with over-training and on downstream tasks

GPQA: A Graduate-Level Google-Proof Q&A Benchmark

JetStream is a throughput and memory optimized engine for LLM inference on XLA devices, starting with TPUs (and GPUs in future -- PRs welcome).

Grok open release

Accelerate your training with this open-source library. Optimize performance with streamlined training and serving options with JAX. 🚀

SGLang is a structured generation language designed for large language models (LLMs). It makes your interaction with models faster and more controllable.

The TinyLlama project is an open endeavor to pretrain a 1.1B Llama model on 3 trillion tokens.

Fetches random solutions to a specified Advent of Code puzzle from GitHub and runs them against your input

MLX: An array framework for Apple silicon

Consistency Distilled Diff VAE

Exploring lessons of mathematics through NFL Scorigami