refactor mm training

Former-commit-id: 179c0558699e287cbf38a2d73bff47e86d589c5a

examples/train_full/qwen2vl_full_sft.yaml

@@ -1,40 +0,0 @@
-### model
-model_name_or_path: qwen2-vl-hf/qwen2-vl-7b-hf
-visual_inputs: true
-
-### method
-stage: sft
-do_train: true
-finetuning_type: full
-deepspeed: examples/deepspeed/ds_z3_config.json
-
-### dataset
-dataset: qwen2vl_demo
-template: qwen2vl
-cutoff_len: 1024
-max_samples: 1000
-overwrite_cache: true
-preprocessing_num_workers: 16
-
-### output
-output_dir: saves/qwen2-vl-7b/full/sft
-logging_steps: 10
-save_steps: 500
-plot_loss: true
-overwrite_output_dir: true
-
-### train
-per_device_train_batch_size: 1
-gradient_accumulation_steps: 1
-learning_rate: 1.0e-5
-num_train_epochs: 3.0
-lr_scheduler_type: cosine
-warmup_ratio: 0.1
-bf16: true
-ddp_timeout: 180000000
-
-### eval
-val_size: 0.1
-per_device_eval_batch_size: 1
-eval_strategy: steps
-eval_steps: 500