You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I am trying to fine tuning Bert on my custom dataset. Training works fine on GPU, however, I am unable to run the training on Google's colab with TPU. I see following error when running with use_tpu=true. My run the script with following params:
I am trying to fine tuning Bert on my custom dataset. Training works fine on GPU, however, I am unable to run the training on Google's colab with TPU. I see following error when running with use_tpu=true. My run the script with following params:
!python BERT_NER.py
--task_name="NER"
--do_lower_case=False
--crf=True
--do_train=True
--do_eval=False
--do_predict=True
--data_dir=<google_drive_location>
--do_lower_case=False
--vocab_file=gs://<bucket_name>/vocab.txt
--bert_config_file=gs://<bucket_name>/bert_config.json
--init_checkpoint=gs://<bucket_name>/bert_model.ckpt
--max_seq_length=512
--train_batch_size=16
--learning_rate=2e-5
--num_train_epochs=5
--output_dir=gs://<bucket_name>/bert-ner-output
--use_tpu=true
--tpu_name={TPU_WORKER}
--num_hosts=1
--num_core_per_host=8
The text was updated successfully, but these errors were encountered: