Build AI models for specialized tasks.
Haven gives you tools needed to build specialized large language models.
Our platform lets you to fine-tune LLMs through a simple UI and evaluate them based on a wide range of criteria.
Welcome 🥳
Welcome to the Haven repository! We initially started out building functionality for LLM fine-tuning, but noticed that most people actually struggle with evaluation and data collection rather than model training itself. To address this, we're planning to add the following features soon:
- Evaluation: Use prebuilt or define custom evaluation metrics and run experiments comparing models across them
- Visualization: Visualize and search through your experiments to get insights on model performance
- Data Collection: Collecting and formatting training datasets is annyoing. We don't really know how to solve this yet, but we're trying to figure it out!
If you have feedback or suggestions on these problems, please reach out! You can join our Discord, write us an email, or schedule a call.
Getting Started :rocket:
Instructions to self-host as well as support for AWS, GCP and Azure will follow soon. In the meantime, you can try our hosted app here.