A natural language interface for computers

Unsloth Studio is a web UI for training and running open models like Qwen3.5, Gemma 4, DeepSeek, gpt-oss locally.

Composable transformations of Python+NumPy programs: differentiate, vectorize, JIT to GPU/TPU, and more

A generative world for general-purpose robotics & embodied AI learning.

[NeurIPS'23 Oral] Visual Instruction Tuning (LLaVA) built towards GPT-4V level capabilities and beyond.

A reactive notebook for Python — run reproducible experiments, query with SQL, execute as a script, deploy as an app, and version with git. Stored as pure Python. All in a modern, AI-native editor.

🦉 OWL: Optimized Workforce Learning for General Multi-Agent Assistance in Real-World Task Automation

A TTS model capable of generating ultra-realistic dialogue in one pass.

Mamba SSM architecture

Semantic segmentation models with 500+ pretrained convolutional and transformer-based backbones.

An open source multi-tool for exploring and publishing data

Open source annotation tool for machine learning practitioners.

Flexible and powerful tensor operations for readable and reliable code (for pytorch, jax, TF and others)

Simple and efficient pytorch-native transformer text generation in <1000 LOC of python.

A concise but complete full-attention transformer with a set of promising experimental features from various papers

A PyTorch library for implementing flow matching algorithms, featuring continuous and discrete flow matching implementations. It includes practical examples for both text and image modalities.

Vector (and Scalar) Quantization, in Pytorch

The fastest ⚡️ way to build data pipelines. Develop iteratively, deploy anywhere. ☁️

PyTorch code and models for VJEPA2 self-supervised learning from video.

Unbearably fast near-real-time pure-Python runtime-static type-checker.

Web-based 3D visualization in Python

[CVPR'25 Oral] MoGe: Unlocking Accurate Monocular Geometry Estimation for Open-Domain Images with Optimal Training Supervision

Code for BLT research paper

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

Continuous Thought Machines, because thought takes time and reasoning is a process.

Implementation of Alphafold 3 from Google Deepmind in Pytorch

Official implementation of "MIMO: Controllable Character Video Synthesis with Spatial Decomposed Modeling"

[ICCV 2025] Implementation for Describe Anything: Detailed Localized Image and Video Captioning

code for "Diffusion Forcing: Next-token Prediction Meets Full-Sequence Diffusion"

Adversarial skill embeddings for training reusable controllers for physically simulated characters.