paulwong16 / evalplus

EvalPlus for rigourous evaluation of LLM-synthesized code

EvalPlus(📖) => 📚

Warning

🚨 Evaluating LLM-generated code on a dataset with just _3_ test-cases is NOT enough! 🚨

To address this, we started the EvalPlus project -- a rigourous evaluation framework for LLM4Code that:

• ✨ improves programming benchmarks by patching up to thousands of new tests! EvalPlus(HumanEval) => HumanEval+ (81x new tests!)
• ✨ crafts a set utility tools to sanitize, visualize and inspect LLM-generated code and evaluation results!
• ✨ accelerates LLM4Code research by open-sourcing LLM-generated samples for 14+ models -- no need to re-run the expensive benchmarks!

Read our paper for more detailed findings!

Use EvalPlus-enhanced dataset

To get started, please first setup the environment:

git clone https://github.com/evalplus/evalplus.git
cd evalplus
pip install -r requirements.txt
export PYTHONPATH=$PYTHONPATH:$(pwd)

HumanEval+

from evalplus.data import get_human_eval_plus

fe = get_human_eval_plus() # -> a list of dictionaries (each is a programming problem)
# "task_id" is the identifier string for the task
# "entry_point": name of the function
# "prompt" is the function signature with docstring
# + "canonical_solution" is the ground-truth implementation (re-implemented to fix bugs in HumanEval)
# + "base_input" is the test inputs in original HumanEval
# + "plus_input" is the test inputs brought by EvalPlus
# and others...

MBPP+ (TBD)

Useful tools

Syntax checker for LLM-generated code

Check LLM-produced code and answer the following questions:

1. Is the generation entirely done for all samples / all problems in the dataset?
2. Are LLM-generated code compilable? (if no, something could be wrong and you'd better check)

python tools/checker.py --folder /path/to/[model]-[??]b_temp_[??] --dataset humaneval

Post code sanitizer

LLM-generated code may contain some syntax errors. But some of them can be easily fixable by doing simple post-processing. This tool will make the LLM-generated code more clean/compilable by doing certain post-processing such as trimming with more magical EOFs and some garbage non-code tokens.

python tools/sanitize.py --eof --folder /path/to/vicuna-[??]b_temp_[??]
# Sanitized code will be produced to `/path/to/vicuna-[??]b_temp_[??]-sanitized`

Render pass@k results to rich and LaTeX tables

python tools/render.py --type /path/to/[model]-[??]b # NOTE: no `_temp_[??]`

Perform test input generation from scratch (TBD)

Development

Before you start:

pip install -r requirements.txt
pre-commit install
export PYTHONPATH=$PYTHONPATH:$(pwd)

Name convention

• evalplus is the package name.
• ${DATASET}_plus is the name of dataset applied with evalplus.

Citation

@article{evalplus,
  title={Is Your Code Generated by ChatGPT Really Correct? Rigorous Evaluation of Large Language Models for Code Generation},
  author={Jiawei Liu and Chunqiu Steven Xia and Yuyao Wang and Lingming Zhang},
  journal={arXiv preprint arXiv:2305.01210},
  year={2023},
}

Acknowledgement

• HumanEval