sft+freeze训练internlm2-base-7b报错，RuntimeError: element 0 of tensors does not require grad and does not have a grad_fn

[1737686924]

Reminder

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

System Info

ASCEND_RT_VISIBLE_DEVICES=0,1 deepspeed --num_gpus 2 src/train.py
--deepspeed examples/deepspeed/ds_z3_offload_config.json
--stage sft
--do_train true
--model_name_or_path /data/applications/lmd-formal/backend/BaseModels/internlm2-base-7b
--dataset identity,alpaca_en_demo
--template intern2
--finetuning_type freeze
--freeze_trainable_layers 8
--freeze_trainable_modules all
--use_llama_pro true
--output_dir saves/internlm2-base-7b/sft/freeze
--overwrite_cache
--overwrite_output_dir
--cutoff_len 1024
--preprocessing_num_workers 16
--per_device_train_batch_size 2
--per_device_eval_batch_size 2
--gradient_accumulation_steps 4
--lr_scheduler_type cosine
--logging_steps 1
--save_steps 100
--eval_steps 100
--evaluation_strategy steps
--load_best_model_at_end
--learning_rate 1e-4
--num_train_epochs 3.0
--val_size 0.001
--ddp_timeout 180000000
--plot_loss
--fp16

sft+freeze训练internlm2-base-7b报错，RuntimeError: element 0 of tensors does not require grad and does not have a grad_fn

transformers=0.41.2
torch=2.2.0
torch-npu=2.2.0

Reproduction

sft+freeze训练internlm2-base-7b报错，RuntimeError: element 0 of tensors does not require grad and does not have a grad_fn

Expected behavior

No response

Others

No response