UserWarning: None of the inputs have requires_grad=True. Gradients will be None

[wangchunhui06]

Reminder

• I have read the README and searched the existing issues.

System Info

_env/lib/python3.10/site-packages/torch/utils/checkpoint.py:87: UserWarning: None of the inputs have requires_grad=True. Gradients will be None

Reproduction

CUDA_VISIBLE_DEVICES=0 llamafactory-cli train
--model_name_or_path /model/llama_3.2v_11B
--stage sft
--do_train true
--finetuning_type full
--freeze_vision_tower true
--train_mm_proj_only false
--dataset llava_cot_train
--template mllama
--cutoff_len 2048
--max_samples 1000
--overwrite_cache true
--preprocessing_num_workers 16
--output_dir /saves/llama_3.2v_11B/full/sft
--logging_steps 10
--save_steps 500
--plot_loss true
--overwrite_output_dir true
--per_device_train_batch_size 4
--gradient_accumulation_steps 1
--learning_rate 1.0e-5
--num_train_epochs 3
--lr_scheduler_type cosine
--warmup_ratio 0.1
--bf16 true
--ddp_timeout 180000000
--val_size 0.1
--per_device_eval_batch_size 1
--eval_strategy steps
--eval_steps 500

Expected behavior

解决问题

Others

无

[hiyouga]

UserWarning 不是报错

[wangchunhui06]

好滴 谢谢