search

huggingface / transformers

🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.
https://huggingface.co/transformers
Apache License 2.0
135.01k stars 27.01k forks source link

transformers_4.13.devo giving error during saving model #14304

Closed Aniruddha-JU closed 2 years ago

Aniruddha-JU commented 3 years ago

Environment info

@patil-suraj Model :mt5-base

input : python run_summarization.py --model_name_or_path google/mt5-base --do_train --do_predict --train_file /home/aniruddha/mt5_data/bengali_8_shot.json --test_file /home/aniruddha/mt5_data/ben_dev.json --source_prefix "summarize: " --output_dir mt5_ben_16_667/ --overwrite_output_dir --per_device_train_batch_size=1 --per_device_eval_batch_size=4 --predict_with_generate --seed 667 --save_steps 14000000 --num_beams 3

error:Training completed. Do not forget to share your model on huggingface.co/models =)

{'train_runtime': 13.5967, 'train_samples_per_second': 1.765, 'train_steps_per_second': 0.883, 'train_loss': 12.063149770100912, 'epoch': 3.0} 100%|████████████████████████████████████████████████████████████████████████████████████████████████████| 12/12 [00:13<00:00, 1.13s/it] [INFO|trainer.py:1995] 2021-11-06 20:20:47,931 >> Saving model checkpoint to mt5_ben_16_667/ [INFO|configuration_utils.py:417] 2021-11-06 20:20:47,932 >> Configuration saved in mt5_ben_16_667/config.json Traceback (most recent call last): File "run_summarization.py", line 648, in main() File "run_summarization.py", line 571, in main trainer.save_model() # Saves the tokenizer too for easy upload File "/home/aniruddha/anaconda3/envs/ani/lib/python3.8/site-packages/transformers/trainer.py", line 1961, in save_model self._save(output_dir) File "/home/aniruddha/anaconda3/envs/ani/lib/python3.8/site-packages/transformers/trainer.py", line 2009, in _save self.model.save_pretrained(output_dir, state_dict=state_dict) File "/home/aniruddha/anaconda3/envs/ani/lib/python3.8/site-packages/transformers/modeling_utils.py", line 1053, in save_pretrained del state_dict[ignore_key] KeyError: 'encoder\.embed_tokens\.weight' Capture

github-actions[bot] commented 2 years ago

This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.

Please note that issues that do not follow the contributing guidelines are likely to be ignored.