train_rm.sh 357 Bytes
Newer Older
1
set_n_least_used_CUDA_VISIBLE_DEVICES 1
ver217's avatar
ver217 committed
2

3
4
5
6
7
8
python train_reward_model.py --pretrain '/home/lczht/data2/bloom-560m' \
                             --model 'bloom' \
                             --strategy naive \
                             --loss_fn 'log_exp'\
                             --save_path 'rmstatic.pt' \
                             --test True