yuweiyin / llm-math-analysis

LLM-Math-Analysis

LLM-Math-Analysis

LLM-Math-Analysis

Environment

conda create -n math -y python=3.9
conda activate math
pip install -r requirements.txt

Common Crawl - Text-Math Extraction

bash run_cc_data.sh
python3 merge_cc_data.py
bash run_cc_analyze.sh
python3 merge_cc_data.py --merge_clear

Deepmind Mathematics Dataset Generation

bash generate_split.sh

StackExchange Text-Math Data

cd ./RedPajama-Data/data_prep/stack_exchange/
python3 download.py

Tsinghua MathGLM

• https://github.com/THUDM/MathGLM

For arithmetic tasks, please download pre-training dataset from MathGLM-dataset. For amth word problem, the reconstructed Ape210K dataset is provided in MathGLM_MWP/dataset/data.jsonl

GSM8K

• https://paperswithcode.com/dataset/gsm8k
• https://github.com/openai/grade-school-math

Berkeley MATH

• https://paperswithcode.com/dataset/math
• https://github.com/hendrycks/math/

MMLU (Massive Multitask Language Understanding)

• https://paperswithcode.com/dataset/mmlu
• https://github.com/hendrycks/test

---