[Question]: ernie-3-tiny 按文档步骤运行例子报错

[LiSongRan]

请提出你的问题

https://github.com/PaddlePaddle/PaddleNLP/blob/71bfc333ade82abdb05f0348ff00ba00cf3c73d9/slm/model_zoo/ernie-3.0-tiny/README.md

python run_train.py
--device gpu
--logging_steps 100
--save_steps 100
--eval_steps 100
--model_name_or_path ernie-3.0-tiny-nano-v2-zh
--num_train_epochs 30
--per_device_eval_batch_size 64
--per_device_train_batch_size 64
--learning_rate 5e-5
--prune_embeddings
--max_vocab_size 6000
--max_seq_length 16
--output_dir output/BS64_LR5e-5_EPOCHS30
--train_path data/train.txt
--dev_path data/dev.txt
--intent_label_path data/intent_label.txt
--slot_label_path data/slot_label.txt
--label_names 'intent_label' 'slot_label'
--weight_decay 0.01
--warmup_ratio 0.1
--do_train
--do_eval
--do_export
--input_dtype "int32"
--disable_tqdm True
--overwrite_output_dir
--load_best_model_at_end True
--save_total_limit 1
--metric_for_best_model eval_accuracy \

案例子运行报错 报错为：
Traceback (most recent call last):
File "/home2/mns_python_sdk/sequence_ner/PaddleNLP/slm/model_zoo/ernie-3.0-tiny/run_train.py", line 286, in
main()
File "/home2/mns_python_sdk/sequence_ner/PaddleNLP/slm/model_zoo/ernie-3.0-tiny/run_train.py", line 151, in main
model = JointErnie.from_pretrained(
File "/root/anaconda3/envs/py310_sr/lib/python3.10/site-packages/paddlenlp/transformers/model_utils.py", line 2529, in from_pretrained
model, missing_keys, unexpected_keys, mismatched_keys = cls._load_pretrained_model(
File "/root/anaconda3/envs/py310_sr/lib/python3.10/site-packages/paddlenlp/transformers/model_utils.py", line 2216, in _load_pretrained_model
raise RuntimeError(f"Error(s) in loading state_dict for {model.class.name}:\n\t{error_msg}")
RuntimeError: Error(s) in loading state_dict for JointErnie:
Skip loading for ernie.embeddings.word_embeddings.weight. ernie.embeddings.word_embeddings.weight receives a shape [5965, 312], but the expected shape is [6000, 312].
You may consider adding ignore_mismatched_sizes=True in the model from_pretrained method.