bf16: true
cutoff_len: 1024
dataset: RPGPT_PublicDomain_v2_alpaca,Footjob
dataset_dir: /workspace/LLaMA-Factory/data/
ddp_timeout: 180000000
do_train: true
finetuning_type: lora
flash_attn: auto
gradient_accumulation_steps: 8
learning_rate: 5.0e-05
logging_steps: 5
lora_alpha: 16
lora_dropout: 0
lora_rank: 8
lora_target: all
lr_scheduler_type: cosine
max_grad_norm: 1.0
max_samples: 100000
model_name_or_path: /workspace/ablation/Qwen1.5-14B-Chat-NSFW/
num_train_epochs: 5.0
optim: adamw_torch
output_dir: saves/Qwen1.5-14B-Chat/lora/Qwen1.5_14B_ChatNSFW_RPGPT+FootJobE5V1
packing: false
per_device_train_batch_size: 2
plot_loss: true
preprocessing_num_workers: 16
report_to: none
save_steps: 100
stage: sft
template: qwen
warmup_steps: 0