forked from harvardnlp/cascaded-generation
-
Notifications
You must be signed in to change notification settings - Fork 0
/
6_cmd.sh
2 lines (2 loc) · 866 Bytes
/
6_cmd.sh
1
2
CUDA_VISIBLE_DEVICES=2 fairseq-generate ../fairseq_vanilla/data-bin/wmt16_en_ro_joint --path /n/rush_lab/users/y/checkpoints/barrier/en-ro-mlm-3gpu-largelr-fix/checkpoint97.pt --batch-size 1 --topk 32 --rounds 4 --remove-bpe --D 3 --max-len-a 0.996921062198531 --max-len-b 1.0282739671455445 --gen-subset valid --max-size 3000 --seed 1234 > /n/rush_lab/users/y/val_ab_logs/wmt-en-ro-epoch-ab-fixD0/topk32_D3_rounds4_speed.txt 2>&1
CUDA_VISIBLE_DEVICES=2 fairseq-generate ../fairseq_vanilla/data-bin/wmt16_en_ro_joint --path /n/rush_lab/users/y/checkpoints/barrier/en-ro-mlm-3gpu-largelr-fix/checkpoint97.pt --batch-size 1 --topk 128 --rounds 4 --remove-bpe --D 3 --max-len-a 0.996921062198531 --max-len-b 1.0282739671455445 --gen-subset valid --max-size 3000 --seed 1234 > /n/rush_lab/users/y/val_ab_logs/wmt-en-ro-epoch-ab-fixD0/topk128_D3_rounds4_speed.txt 2>&1