A Simple Postgres Logger for OpenAI Endpoints

Learn how to implement a simple Postgres logger for OpenAI endpoints in this 19-minute tutorial from Trelis Research. Discover why logging is essential for evaluations and fine-tuning, and explore the open-source trelis-openai-logger package available on GitHub and PyPI. The tutorial demonstrates how to log API traces to Postgres databases in three different environments: locally, on a remote server (Digital Ocean Droplet), and using Digital Ocean's Managed Database service. The video includes a practical demonstration of the logger in action and previews future content on using Postgres logs for running evaluations and creating fine-tuning datasets. Install with "pip install trelis-openai-logger" and start tracking your OpenAI API interactions efficiently.

Syllabus

0:00 Why logging is useful for evals and fine-tuning
0:52 Finding the Github Repo and PyPI package: https://github.com/TrelisResearch/trelis-openai-logger
1:22 Quick demo of trelis-openai-logger - logging traces to Postgres
4:26 Running Postgres locally
8:36 Running Postgres on a remote server Digital Ocean Droplet
15:31 Running Postgres as a Digital Ocean Managed Database
17:36 Future Video: Using Postgres logs to run evals and create fine-tuning datasets