Komiko - Create comics, manhwa, manga, webtoon, and anime with AI - AI Comic Factory

Finetune Llama 4, DeepSeek-R1, Gemma 3 & Reasoning LLMs 2x faster with 70% less memory! 🦥

[CVPR 2025 Oral]Infinity ∞ : Scaling Bitwise AutoRegressive Modeling for High-Resolution Image Synthesis

[SIGGRAPH Asia 2024, Best Paper Honorable Mention] This is the official implementation of our SIGGRAPH Asia journal artical: TEXGen: a Generative Diffusion Model for Mesh Textures

Material for gpu-mode lectures

Efficient Triton Kernels for LLM Training

Official repository of In-Context LoRA for Diffusion Transformers

$100K or 100 Days: Trade-offs when Pre-Training with Academic Resources

[ICLR2025 Spotlight🔥] Official Implementation of TokenFormer: Rethinking Transformer Scaling with Tokenized Model Parameters

[NeurIPS 2024] DreamClear: High-Capacity Real-World Image Restoration with Privacy-Safe Dataset Curation

OpenR: An Open Source Framework for Advanced Reasoning with Large Language Models

Lightweight Python framework that provides a high-level API for creating and rendering scenes with Blender.

Fira: Can We Achieve Full-rank Training of LLMs Under Low-rank Constraint?

[ICLR'25 Oral] Representation Alignment for Generation: Training Diffusion Transformers Is Easier Than You Think

Mirage: Automatically Generating Fast GPU Kernels without Programming in Triton/CUDA

[CVPR 2025] 3DTopia-XL: High-Quality 3D PBR Asset Generation via Primitive Diffusion

A python library for self-supervised learning on images.

Distributed Training Over-The-Internet

A compact LLM pretrained in 9 days by using high quality data

🚀 Efficient implementations of state-of-the-art linear attention models in Torch and Triton

Lighteval is your all-in-one toolkit for evaluating LLMs across multiple backends

Freeing data processing from scripting madness by providing a set of platform-agnostic customizable pipeline processing blocks.

This is the official repository for Inheritune.

[ICLR 2025] Mathematical Visual Instruction Tuning for Multi-modal Large Language Models

ControlNet++: All-in-one ControlNet for image generations and editing!

MINT-1T: A one trillion token multimodal interleaved dataset.

[ICLR 2025 Spotlight] OmniCorpus: A Unified Multimodal Corpus of 10 Billion-Level Images Interleaved with Text