Differential transformers implementation and axolotl plugin

Official repository for the paper "Grokfast: Accelerated Grokking by Amplifying Slow Gradients"

Official implementation for DenseMixer: Improving MoE Post-Training with Precise Router Gradient

Codebase/framework for training Large Language Model.

DeepSpeed is a deep learning optimization library that makes distributed training and inference easy, efficient, and effective.

Environments for LLM Reinforcement Learning

A PyTorch native platform for training generative AI models

An advanced singing voice synthesis system with high fidelity, expressiveness, controllability and flexibility based on DiffSinger: Singing Voice Synthesis via Shallow Diffusion Mechanism

OneTrainer is a one-stop solution for all your stable diffusion training needs.

Foundation Architecture for (M)LLMs

色々なカスタムノードを詰め込んだ。

An easy to use loot generator app designed for Advanced Dungeons & Dragons 2nd edition (AD&D 2e), but flexible enough to be adapted to other systems.

Translate manga/image 一键翻译各类图片内文字 https://cotrans.touhou.ai/ (no longer working)

Git Re-Basin: Merging Models modulo Permutation Symmetries in PyTorch

A fast CUDA implementation of the Linear Assignment Problem (LAP) solver for PyTorch.

An enhanced version of Tomas Kazmar's lap — Linear Assignment Problem solver (LAPJV/LAPMOD).

Adaptation of the merging method described in the paper - Git Re-Basin: Merging Models modulo Permutation Symmetries (https://arxiv.org/abs/2209.04836) for Stable Diffusion

Adaptation of the merging method described in the paper - Git Re-Basin: Merging Models modulo Permutation Symmetries (https://arxiv.org/abs/2209.04836) for Stable Diffusion

This repository is the implementation of the paper Training Free Pretrained Model Merging (CVPR2024).

Code release for "Git Re-Basin: Merging Models modulo Permutation Symmetries"

[ECCV 2024] official code for "Long-CLIP: Unlocking the Long-Text Capability of CLIP"

Repo for "STAR: Spectral Truncation and Rescale for Model Merging" [NAACL 2025]

A 100x faster SVD for PyTorch⚡️

[ICML 2025] No Task Left Behind: Isotropic Model Merging with Common and Task-Specific Subspaces (official repository)

Open source free capture HTTP(S) traffic software ProxyPin, supporting full platform systems