An open-weights time-series forecasting foundation model from The Forecasting Company.

An LLM post-training framework with vLLM for RL Scaling

Model export recipes, Python primitives, and Swift runtime utilities for on-device AI

Triton kernels for dynamic causal short convolutions.

Context Parallelism utilities for Training Language Models

Native macOS semantic search over your local files - text, images, audio, video in one vector space, on-device on Apple silicon.

AutoScientists: Self-Organizing Agent Teams for Long-Running Scientific Experimentation

CRANE: Cluster-Reactive Adaptive News Ensemble — A CPU-native sentiment engine that reads news, predicts markets, and adapts to regime shifts without a GPU.

torch_remat fine-grained activation checkpointing API

TokenSpeed is a speed-of-light LLM inference engine.

Open ABI and FFI for Machine Learning Systems

GEMMs with metal

Convert any Repo into an RL Environment

Code for Retrofitting Large Language Models with Dynamic Tokenization.

DiffusionBlocks: Block-wise Neural Network Training via Diffusion Interpretation

mKernel: fast multi-node, multi-GPU fused kernels

A unified library for building, evaluating, and storing speculative decoding algorithms for LLM inference in vLLM

Train speculative decoding models effortlessly and port them smoothly to SGLang serving.

26m function call model that runs on incredibly small devices

Open-source framework for the research and development of foundation models.

Personal dev tool that exposes a local shell and filesystem to any Model Context Protocol client. Built on the official TypeScript SDK, with a custom WebSocket server transport.

The home of Carbon Genomic Foundation Model 🧬

🤗 ml-intern: an open-source ML engineer that reads papers, trains models, and ships ML models

Simple & Scalable Pretraining for Neural Architecture Research