evaluation/general_benchmarks/MATH/README.md

Requirements

You can install the required packages with the following command:

cd latex2sympy
pip install -e .
cd ..
pip install -r requirements.txt 
pip install vllm==0.5.1 --no-build-isolation
pip install transformers==4.42.3

Evaluation

You can evaluate Qwen2.5/Qwen2-Math-Instruct series model with the following command:

# Qwen2.5-Math-Instruct Series
PROMPT_TYPE="qwen25-math-cot"
# Qwen2.5-Math-1.5B-Instruct
export CUDA_VISIBLE_DEVICES="0"
MODEL_NAME_OR_PATH="Qwen/Qwen2.5-Math-1.5B-Instruct"
bash sh/eval.sh $PROMPT_TYPE $MODEL_NAME_OR_PATH

# Qwen2.5-Math-7B-Instruct
export CUDA_VISIBLE_DEVICES="0"
MODEL_NAME_OR_PATH="Qwen/Qwen2.5-Math-7B-Instruct"
bash sh/eval.sh $PROMPT_TYPE $MODEL_NAME_OR_PATH

# Qwen2.5-Math-72B-Instruct
export CUDA_VISIBLE_DEVICES="0,1,2,3"
MODEL_NAME_OR_PATH="Qwen/Qwen2.5-Math-72B-Instruct"
bash sh/eval.sh $PROMPT_TYPE $MODEL_NAME_OR_PATH


# Qwen2-Math-Instruct Series
PROMPT_TYPE="qwen-boxed"
# Qwen2-Math-1.5B-Instruct
export CUDA_VISIBLE_DEVICES="0"
MODEL_NAME_OR_PATH="Qwen/Qwen2-Math-1.5B-Instruct"
bash sh/eval.sh $PROMPT_TYPE $MODEL_NAME_OR_PATH

# Qwen2-Math-7B-Instruct
export CUDA_VISIBLE_DEVICES="0"
MODEL_NAME_OR_PATH="Qwen/Qwen2-Math-7B-Instruct"
bash sh/eval.sh $PROMPT_TYPE $MODEL_NAME_OR_PATH

# Qwen2-Math-72B-Instruct
export CUDA_VISIBLE_DEVICES="0,1,2,3"
MODEL_NAME_OR_PATH="Qwen/Qwen2-Math-72B-Instruct"
bash sh/eval.sh $PROMPT_TYPE $MODEL_NAME_OR_PATH

Acknowledgement

The codebase is adapted from math-evaluation-harness.