llama-factory/config/mac_template_4gpu.yaml

### model
model_name_or_path: ORG_NAME/MODEL_NAME
quantization_bit: 4

### method
stage: sft
do_train: true
finetuning_type: lora
lora_target: all

### dataset
dataset: alpaca_mac
template: CHAT_TEMPLATE
cutoff_len: 1024
max_samples: 4528
overwrite_cache: true
preprocessing_num_workers: 16

### output
output_dir: saves/MODEL_NAME
logging_steps: 5
save_steps: 70
plot_loss: true
# overwrite_output_dir: true

### train
per_device_train_batch_size: 2
gradient_accumulation_steps: 8
learning_rate: 1.0e-4
num_train_epochs: 6.0
lr_scheduler_type: cosine
warmup_ratio: 0.1
bf16: true
ddp_timeout: 180000000

### eval
val_size: 0.01
per_device_eval_batch_size: 1
eval_strategy: steps
eval_steps: 70

report_to: wandb
run_name: MODEL_NAME_lora_sft