Compress and Attend Transformers (CATs) 😸

puffer_drive dev

Hand-Rolled GPU communications library

Large multi-modal models (L3M) pre-training.

CLI to manage emails

1 million FPS multi-agent driving simulator

Stable Diffusion web UI

The definitive Web UI for local AI, with powerful features and easy setup.

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

Minimal pretraining script for language modeling in PyTorch. Supporting torch compilation and DDP. It includes a model implementation and a data preprocessing script.

Inference-time scaling for LLMs-as-a-judge.

🌾 OAT: A research-friendly framework for LLM online alignment, including reinforcement learning, preference learning, etc.

A Survey of Reinforcement Learning for Large Reasoning Models

A modular, primitive-first, python-first PyTorch library for Reinforcement Learning.

Lightweight coding agent that runs in your terminal

M1: Towards Scalable Test-Time Compute with Mamba Reasoning Models

Adaptation of titans-pytorch to llama models on HF

Unofficial implementation of Titans, SOTA memory for transformers, in Pytorch

Production-tested AI infrastructure tools for efficient AGI development and community-driven innovation

Repo for "LoLCATs: On Low-Rank Linearizing of Large Language Models"

🔥 A minimal training framework for scaling FLA models

Fully open reproduction of DeepSeek-R1

A system monitoring tool powered by LLMs that provides real-time insights about your system's performance

The official repo of MiniMax-Text-01 and MiniMax-VL-01, large-language-model & vision-language-model based on Linear Attention

🚀 Efficient implementations of state-of-the-art linear attention models

AlphaFold 3 inference pipeline.