ESFT/scripts/eval.sh

25 lines
897 B
Bash
Raw Normal View History

2024-08-09 10:06:57 +00:00
# first: download adapter models and put them to the corresponding directories
2024-08-09 10:06:57 +00:00
python eval_multigpu.py \
--eval_datasets=translation \
--base_model_path=deepseek-ai/ESFT-vanilla-lite \
--adapter_dir=all_models/adapters/token \
--output_dir=results/completions/token \
--max_new_tokens=512 \
--openai_api_key=REPLACE_WITH_YOUR_KEY \
2024-08-09 10:06:57 +00:00
--eval_batch_size=2 \
--world_size=4 \
--gpus_per_rank=2
2024-08-09 10:06:57 +00:00
# this script is used for single-gpu training and has been deprecated. If you have no multiple gpus, you can set above world_size=1 and gpus_per_rank=1
# python scripts/eval.py \
# --eval_datasets=translation \
# --base_model_path=deepseek-ai/ESFT-vanilla-lite \
# --adapter_dir=all_models/adapters/token \
# --output_dir=results/completions/token \
# --max_new_tokens=512 \
# --openai_api_key=REPLACE_WITH_YOUR_KEY \
# --eval_batch_size=2