Upload training_args.yaml with huggingface_hub

training_args.yaml

@@ -0,0 +1,29 @@
+bf16: true
+cutoff_len: 2048
+dataset: xiaosui-train,identity
+dataset_dir: data
+ddp_timeout: 180000000
+do_train: true
+finetuning_type: full
+flash_attn: auto
+gradient_accumulation_steps: 8
+learning_rate: 5.0e-05
+logging_steps: 5
+lr_scheduler_type: cosine
+max_grad_norm: 1.0
+max_samples: 5000
+model_name_or_path: Qwen/Qwen2.5-0.5B-Instruct
+num_train_epochs: 3.0
+optim: adamw_torch
+output_dir: saves/Qwen2.5-0.5B-Instruct/full/train_2025-01-23-16-59-22
+packing: false
+per_device_train_batch_size: 2
+plot_loss: true
+preprocessing_num_workers: 16
+report_to:
+- wandb
+save_steps: 100
+stage: sft
+template: qwen
+trust_remote_code: true
+warmup_steps: 4