Welcome to Predibase

Predibase is a low-code AI platform built for developers. Train, finetune, and deploy any model, from linear regression to large language models. Predibase is built on top of open source library, Ludwig.

Two ways to get started

1. Large Language Models

Try out popular open source LLMs. Predibase enables you to do zero and few shot prompting, dataset indexing, and batch inference using LLMs. After testing out different LLMs and prompts, you can also customize an LLM via fine-tuning on a custom dataset or evaluate an LLM's abiity to perform on a specific task and generate metrics.

Check out our LLM quickstart guide.

2. Supervised ML

If you have specific task in mind and a labeled dataset, you can train a deep learning or tree model in just a few steps. Connect your dataset, kick off model training, and analyze the results by looking at performance metrics or running a PREDICT PQL query. You can easily compare different model iterations and keep them orgnanized in a model repository.

Check out our supervised ML quickstart guide.

Contact Us

Reach out to us via the chat feature or support@predibase.com for any questions, comments, or concerns.