### model
model_name_or_path: mistralai/Ministral-8B-Instruct-2410
trust_remote_code: true

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

### dataset
dataset: course-train-fold1
dataset_dir: data_private
template: mistral
cutoff_len: 1024
# max_samples: 1000
overwrite_cache: true
preprocessing_num_workers: 16

### output
output_dir: saves/psy-course/Ministral-8B-Instruct-2410/train/fold3
logging_steps: 10
save_steps: 50
plot_loss: true
overwrite_output_dir: true
save_total_limit: 3
push_to_hub: true
hub_model_id: chchen/Ministral-8B-Instruct-2410-PsyCourse-fold3
load_best_model_at_end: true

### train
per_device_train_batch_size: 1
gradient_accumulation_steps: 16
learning_rate: 1.0e-4
num_train_epochs: 5.0
lr_scheduler_type: cosine
warmup_ratio: 0.1
bf16: true
ddp_timeout: 180000000
enable_liger_kernel: true

### eval
val_size: 0.1
per_device_eval_batch_size: 1
eval_strategy: steps
eval_steps: 50