wehub-resource-sync
24 lines
681 B
Bash
24 lines
681 B
Bash
# 4*80G
|
|
# exp: https://github.com/modelscope/ms-swift/pull/5355
|
|
CUDA_VISIBLE_DEVICES=0,1,2,3 \
|
|
NPROC_PER_NODE=4 \
|
|
swift sft \
|
|
--model Qwen/Qwen2.5-Math-1.5B \
|
|
--tuner_type full \
|
|
--dataset AI-MO/NuminaMath-CoT#100000 \
|
|
--load_from_cache_file true \
|
|
--torch_dtype bfloat16 \
|
|
--enable_dft_loss true \
|
|
--num_train_epochs 1 \
|
|
--per_device_train_batch_size 8 \
|
|
--learning_rate 5e-5 \
|
|
--gradient_accumulation_steps 32 \
|
|
--save_total_limit 2 \
|
|
--logging_steps 5 \
|
|
--max_length 2048 \
|
|
--output_dir output \
|
|
--system 'You are a helpful assistant.' \
|
|
--warmup_ratio 0.1 \
|
|
--deepspeed zero2 \
|
|
--dataloader_num_workers 4
|