Skip to content

Latest commit

 

History

History
15 lines (12 loc) · 399 Bytes

README.md

File metadata and controls

15 lines (12 loc) · 399 Bytes

HumanEval for GPT-3.5/GPT-4

Forked from OpenAI's repo. Should run out of the box with python 3.11 (as of 11/23)

To generate the completions (after pip installing the requirements), run:

mkdir results
python run.py

Then to evaluate the completion results, run

python human_eval/evaluate_functional_correctness "YOUR RESULTS FILE.jsonl"