Repolode
Submit
h2oai

h2oai/h2o-llmstudio

A framework and no-code GUI for fine-tuning large language models, with LoRA, DPO and CLI support.

View on GitHubWebsite
Stars
5,193
Forks
555
Open issues
38
Last commit
6 hours ago

What is it?

H2O LLM Studio is a framework and graphical interface for fine-tuning large language models without writing code. Users upload a dataset, choose a base model and set hyperparameters through the GUI, then monitor, compare and chat with the resulting experiments.

The project supports parameter-efficient methods such as Low-Rank Adaptation (LoRA) and 8-bit training to reduce GPU memory use. It also includes preference optimization (DPO, IPO and KTO) and problem types for causal classification and regression, with experimental reinforcement learning support. Training can be driven from a YAML configuration via the command line, and DeepSpeed and multi-GPU DDP are available for larger models.

Trained models can be pushed to the Hugging Face Hub, and Weights & Biases can be used for tracking. The app runs locally on Ubuntu with NVIDIA GPUs, or through Docker and a runpod.io template.

Key features

  • 01No-code GUI for creating, monitoring and comparing fine-tuning experiments.
  • 02LoRA and 8-bit training to lower GPU memory requirements.
  • 03DPO, IPO and KTO preference optimization options.
  • 04Command-line training from YAML configuration files, with multi-GPU DDP support.
  • 05Interactive chat with fine-tuned models and export to the Hugging Face Hub.

Use cases

  • Adapting an open-source LLM to a domain-specific question-answering dataset without writing training code.
  • Comparing hyperparameter settings across several fine-tuning runs in one interface.
  • Running scripted, reproducible fine-tuning jobs on a GPU server through the CLI.

Quick start

  1. 1

    Install the uv-based virtual environment and dependencies

    make setup
  2. 2

    Start the H2O LLM Studio GUI

    make llmstudio
  3. 3

    Open http://localhost:10101/ in your browser

Things to know

  • Requires an Ubuntu machine with a recent NVIDIA GPU and NVIDIA drivers version 580 or newer.
  • For larger models, at least 24GB of GPU memory is recommended.
  • Some features such as RL fine-tuning are marked experimental, and backwards compatibility is not guaranteed.

In collections

Run and train AI on GPUsAI chatbots & chat clients

Topics

#llm#fine-tuning#lora#gpu#generative-ai#no-code#deep-learning#ai#chatbot#chatgpt

FAQ

What is h2o-llmstudio?
A framework and no-code GUI for fine-tuning large language models, with LoRA, DPO and CLI support. H2O LLM Studio is a framework and graphical interface for fine-tuning large language models without writing code. Users upload a dataset, choose a base model and set hyperparameters through the GUI, then monitor, compare and chat with the resulting experiments.
Is h2o-llmstudio free to use? What license does it have?
h2o-llmstudio is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is h2o-llmstudio still maintained?
Actively maintained. The last commit to h2o-llmstudio was on Nov 7, 2026.
What can I use h2o-llmstudio for?
Adapting an open-source LLM to a domain-specific question-answering dataset without writing training code. Comparing hyperparameter settings across several fine-tuning runs in one interface. Running scripted, reproducible fine-tuning jobs on a GPU server through the CLI.
How do I install h2o-llmstudio?
Install the uv-based virtual environment and dependencies: make setup Start the H2O LLM Studio GUI: make llmstudio Open http://localhost:10101/ in your browser
Where can I get h2o-llmstudio?
The source code and installation instructions are on GitHub: https://github.com/h2oai/h2o-llmstudio

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects