Views
No views yet
1WANDB_MODE=disabled python run_grace.py \
2 --model_name_or_path mkhalifa/flan-t5-large-gsm8k \
3 --in_file data/gsm8k/dev.jsonl \
4 --task gsm8k \
5 --disc_path ckpts/discrim/flan-t5-gsm8k/ \
6 --beta 0.1 --n_candidate_steps 20 --generation_type step-score \
7 --step_sampling_method top_p --device2 cuda:0 --top_p .95 --sample_calc true \
8 --max_steps 6 --max_step_length 60 --step_delimiter '|' --temperature .8 --n_self_consistency 1 --seed 421@article{khalifa2023grace,
2 title={Grace: Discriminator-guided chain-of-thought reasoning},
3 author={Khalifa, Muhammad and Logeswaran, Lajanugen and Lee, Moontae and Lee, Honglak and Wang, Lu},
4 journal={arXiv preprint arXiv:2305.14934},
5 year={2023}
6}