이전 코드베이스가 엉망인 관계로 제공함
시작점: https://github.com/fgenie/Model-Selection-Reasoning/tree/59debd8441e7cb4b7d733f256b20870073e36c08
place openai_key.txt
into utils/
pip install pre-commit
pre-commit install
python run_inference.py baseline_inference \
--backbone chatgpt \
--outdir dbgoutdir/ \
--gsm_jslf ../dataset/dbg.jsonl \
--dataset_type gsm # [ocw, math, gsm, svamp]
[--start_idx 0 ] # can start running from the middle of the data
[--dbg] # runs with tqdm instead of pqdm
python run_inference.py rims_inference \
--backbone chatgpt \
--outdir dbgoutdir/ \
--prompt_f prompt_construction_src/prep_rims_prompts/gsm_prompts/3_reflectonce_cot2p2c.pal2cot.pal2p2c.txt_rm_ans \
--gsm_jslf dbgoutdir/chatgpt_01_18_04_48_model_selection3_startidx0.jsonl \
--dataset_type gsm # [ocw, math, gsm, svamp]
# baseline result
python run_evaluation.py --eval_jslf dbgoutdir/chatgpt_01_18_04_48_model_selection3_startidx0.jsonl
# rims result
python run_evaluation.py --eval_jslf dbgoutdir/chatgpt_rims_01_18_04_49_startidx0.jsonl
- add
-
NameError
when running baseline_inference on MATH dataset? - ocw, (math) symbolic including prompts
-
TBA