Linaqruf/kohya-trainer

Train LoRAs on Google Colab in one click: Illustrious/NoobAI, Anima and MOSS-Image, over kohya-ss/sd-scripts.

What it solves

Training LoRAs (Low-Rank Adaptation models) for image generation can be complex and technically demanding. This project simplifies the process by providing ready-to-use Google Colab notebooks that handle the entire training pipeline, ensuring compatibility and stability by pinning to specific versions of the underlying training software.

How it works

The project uses the kohya-ss/sd-scripts backend to perform the actual training. Users can choose between a "1-click" notebook for a streamlined, automated experience or an "advanced" notebook that breaks the process into individual stages (downloading, tagging, configuring, and training) for greater control. The notebooks support various datasets—including local folders, zip files, or Hugging Face repositories—and automatically handle image captioning if needed.

Who it’s for

This tool is designed for users who want to train LoRAs on Google Colab without managing complex local environments. It caters to both beginners who want a simple "press-and-run" experience and power users who need to adjust specific training settings or review intermediate steps.

Highlights

  • One-click training notebooks for popular models like SDXL, SD 1.5, and Illustrious.
  • Advanced mode allows for step-by-step verification and re-running of specific training stages.
  • Native support for uploading results to the Hugging Face Hub and logging progress to Weights & Biases.
  • Flexible dataset handling, including support for repeat counts via subfolder naming and automated tagging.
  • Secure handling of API tokens for Hugging Face, Civitai, and Weights & Biases via Colab Secrets.

Related

  • Project
    stanfordmlgroup/ngboost

    A Python library implementing Natural Gradient Boosting for probabilistic prediction, allowing models to output probability distributions instead of single point estimates.

  • Project
    trevorstephens/gplearn

    A Python library for genetic programming that uses symbolic regression to discover mathematical expressions that describe relationships between variables.

  • Project
    flexflow/flexflow-train

    FlexFlow Train is a deep learning framework that automatically searches for efficient parallelization strategies to accelerate distributed DNN training. It supports PyTorch, Keras, and ONNX models, and uses MCMC-based auto-tuning to find faster execution plans on GPU clusters.

  • Project
    tiepvupsu/ebookMLCB

    This repository provides the source code and PDF versions of the Machine Learning cơ bản textbook, offering a foundational educational resource for students of machine learning.