CUDA on non-NVIDIA GPUs

Fine-tuning & Reinforcement Learning for LLMs. 🦥 Train OpenAI gpt-oss, DeepSeek-R1, Qwen3, Gemma 3, TTS 2x faster with 70% less VRAM.

AI-powered text compression library for RAG systems and API calls. Reduce token usage by up to 50-60% while preserving semantic meaning with advanced compression strategies.

Fully local web research and report writing assistant

Build resilient language agents as graphs.

embedded graph datastore

Lumina-mGPT 2.0: Stand-Alone AutoRegressive Image Modeling

MemU is an open-source memory framework for AI companions

Official inference code and LongText-Bench benchmark for our paper X-Omni (https://arxiv.org/pdf/2507.22058).

Cosmos-Predict2 is a collection of general-purpose world foundation models for Physical AI that can be fine-tuned into customized world models for downstream applications.

Video-based AI memory library. Store millions of text chunks in MP4 files with lightning-fast semantic search. No database needed.

Open-source unified multimodal model

All information and news with respect to Falcon-H1 series

[NeurIPS 2025] MMaDA - Open-Sourced Multimodal Large Diffusion Language Models

For your multi-agent needs

AzerothCore Playerbots Module

Running Stable diffusion from a mobile phone using Termux

[CVPR 2025] Diffusion Self-Distillation for Zero-Shot Customized Image Generation

Implementation of Karpathy's micrograd in Mojo 🔥

A bidirectional pipeline parallelism algorithm for computation-communication overlap in DeepSeek V3/R1 training.

FlashMLA: Efficient Multi-head Latent Attention Kernels

Generative AI extensions for onnxruntime

DroneKit-Python library for communicating with Drones via MAVLink.

Accelerate local LLM inference and finetuning (LLaMA, Mistral, ChatGLM, Qwen, DeepSeek, Mixtral, Gemma, Phi, MiniCPM, Qwen-VL, MiniCPM-V, etc.) on Intel XPU (e.g., local PC with iGPU and NPU, discr…