MongooseMiner is a search system that pushes LLM-based code generation beyond average human performance.
Most LLMs for code generation write code like humans:
- They use the most common packages rather than the most appropriate and powerful ones,
- They use the most common functions instead of the most appropriate and powerful ones,
- They do not memorize all available function arguments and perform many operations where one is sufficient. This is because LLMs learn from the average developer's code, while MongooseMiner learns from the experts who wrote the libraries and documentation.
By evaluating the documentation strings of the most common PyPI projects and retrieving them as needed to guide LLM autocompletion, MongooseMiner can deliver the most appropriate and performant code.
Log in or sign up for Devpost to join the conversation.