-
Notifications
You must be signed in to change notification settings - Fork 279
/
sft.sh
31 lines (31 loc) · 940 Bytes
/
sft.sh
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
# Experimental environment: 4 * A100
# 4 * 60GB GPU memory
# Note: you have to install latest version of the transformers library.
# pip install git+https://github.com/huggingface/transformers
CUDA_VISIBLE_DEVICES=0,1,2,3 \
swift sft \
--model_type c4ai-command-r-plus \
--sft_type lora \
--tuner_backend swift \
--dtype AUTO \
--output_dir output \
--dataset alpaca-zh#10000 alpaca-en#10000 \
--num_train_epochs 2 \
--max_length 2048 \
--check_dataset_strategy warning \
--lora_rank 8 \
--lora_alpha 32 \
--lora_dropout_p 0.05 \
--lora_target_modules DEFAULT \
--gradient_checkpointing true \
--batch_size 1 \
--weight_decay 0.1 \
--learning_rate 1e-4 \
--gradient_accumulation_steps 16 \
--max_grad_norm 0.5 \
--warmup_ratio 0.03 \
--eval_steps 100 \
--save_steps 100 \
--save_total_limit 2 \
--logging_steps 10 \
--use_flash_attn true \