File size: 681 Bytes
bdd5ef6 |
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 |
cutoff_len: 1024
dataset: Central-SheungWan
dataset_dir: data
ddp_timeout: 180000000
do_train: true
finetuning_type: freeze
flash_attn: auto
fp16: true
freeze_trainable_layers: 2
freeze_trainable_modules: all
gradient_accumulation_steps: 8
learning_rate: 5.0e-05
logging_steps: 5
lr_scheduler_type: cosine
max_grad_norm: 1.0
max_samples: 10000
model_name_or_path: shenzhi-wang/Llama3-8B-Chinese-Chat
num_train_epochs: 3.0
optim: adamw_torch
output_dir: saves/LLaMA3-8B-Chinese-Chat/freeze/train_2024-05-30-09-37-42
packing: true
per_device_train_batch_size: 1
plot_loss: true
preprocessing_num_workers: 16
report_to: none
save_steps: 100
stage: pt
template: llama3
warmup_steps: 0
|