p04798526
/
LLaMA-Factory-310P3
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

chore: change lr to original

This commit is contained in:
wql 2024-10-23 16:06:02 +08:00
parent 0dbc298095
commit e6e5cd97e1
2 changed files with 2 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
results/lora_sft_template.yaml
View File

@ -25,7 +25,7 @@ overwrite_output_dir: true
### train ### train
per_device_train_batch_size: 2 per_device_train_batch_size: 2
gradient_accumulation_steps: 8 gradient_accumulation_steps: 8
learning_rate: 1.0e-5 learning_rate: 1.0e-4
num_train_epochs: 10.0 num_train_epochs: 10.0
lr_scheduler_type: cosine lr_scheduler_type: cosine
warmup_ratio: 0.1 warmup_ratio: 0.1

1
run_once.sh
View File

@ -3,6 +3,7 @@
# bash run_once.sh lora_sft Qwen-7B 8 50 # bash run_once.sh lora_sft Qwen-7B 8 50
# bash run_once.sh lora_sft Qwen-7B 1 50 # bash run_once.sh lora_sft Qwen-7B 1 50
# bash run_once.sh lora_sft 9g-8B 8 50
# bash run_once.sh inference Qwen-7B 8 50 # bash run_once.sh inference Qwen-7B 8 50
run_type="$1" run_type="$1"
model="$2" model="$2"