Can I read Evaluating Codex for Code Generation on EtoBox?

Evaluating Codex for Code Generation by jkash is a document available to read on EtoBox.

What is Evaluating Codex for Code Generation about?

This document introduces Codex, an AI model trained by OpenAI to generate Python code. On a new evaluation set called HumanEval that measures functional correctness, Codex solves 28.8% of problems while GPT-3 solves 0% and GPT-J solves 11.4%. Repeated sampling from Codex is an effective strategy, with it solving 70.2% of problems with 100 samples per problem. The document also discusses safety, security and economic impacts of deploying powerful code generation technologies.

Author

jkash

Language

EN