mirror of
https://github.com/deepseek-ai/ESFT
synced 2024-11-22 03:27:38 +00:00
12 lines
432 B
Bash
12 lines
432 B
Bash
|
|
||
|
|
export TOKENIZERS_PARALLELISM=false
|
||
|
|
|
||
|
|
exp_name="test/eval_translation"
|
||
|
|
base_model_path="/hf3fs-jd/prod/deepseek/shared/wangzihan/models/huggingface/vanilla_model"
|
||
|
|
# turn above to for loop
|
||
|
|
python train.py \
|
||
|
|
--base_model_path=${base_model_path} \
|
||
|
|
--expert_config=results/expert_configs/translation.json \
|
||
|
|
--train_dataset=translation \
|
||
|
|
--train_config=configs/base.yaml \
|
||
|
|
--output_dir=results/checkpoints/${exp_name}
|