无法加载 tensorflow BERT 预训练模型

Question

I tried to load a BERT pre-trained model to do NER task. But the system cannot find the pretrained model file.

I used the following code in terminal, the folder contains model.ckpt-1000000, model.ckpt-1000000.index,model.ckpt-1000000.meta files.

python run_ner.py \
    --do_train=true \
    --do_eval=true \
    --vocab_file=vocab.txt \
    --bert_config_file=bert_config.json \
    --init_checkpoint=model.ckpt-1000000 \
    --num_train_epochs=10.0 \
    --data_dir=NCBI-disease/ \
    --output_dir=epoch1

The error message is

2019-08-04 23:26:41.272281: W tensorflow/core/framework/op_kernel.cc:1401] OP_REQUIRES failed at save_restore_v2_ops.cc:184 : Not found: model.ckpt-1000000.data-00000-of-00001; No such file or directory

  File "/Users/anaconda/envs/BIOBERT/lib/python3.6/site-packages/tensorflow/python/client/session.py", line 1328, in _do_run
    run_metadata)
  File "/Users/anaconda/envs/BIOBERT/lib/python3.6/site-packages/tensorflow/python/client/session.py", line 1348, in _do_call
    raise type(e)(node_def, op, message)
tensorflow.python.framework.errors_impl.NotFoundError: model.ckpt-1000000.data-00000-of-00001; No such file or directory
     [[node checkpoint_initializer_161 (defined at run_ner.py:422) ]]

Note: The original names of the model files are model.ckpt-1000000.data-00000-of-00001, model.ckpt-1000000.index and model.ckpt-1000000.meta. I also tried

python run_ner.py \
    --do_train=true \
    --do_eval=true \
    --vocab_file=vocab.txt \
    --bert_config_file=bert_config.json \
    --init_checkpoint=model.ckpt-1000000.data-00000-of-00001 \
    --num_train_epochs=10.0 \
    --data_dir=NCBI-disease/ \
    --output_dir=epoch1

Then the error would be

  File "/Users/anaconda/envs/BIOBERT/lib/python3.6/site-packages/tensorflow/python/pywrap_tensorflow_internal.py", line 326, in NewCheckpointReader
    return CheckpointReader(compat.as_bytes(filepattern), status)
  File "/Users/SichengZhou/anaconda/envs/BIOBERT/lib/python3.6/site-packages/tensorflow/python/framework/errors_impl.py", line 528, in __exit__
    c_api.TF_GetCode(self.status.status))
tensorflow.python.framework.errors_impl.DataLossError: Unable to open table file ./model.ckpt-1000000.data-00000-of-00001: Data loss: not an sstable (bad magic number): perhaps your file is in a different file format and you need to use a different restore operator?

Answer 1

The original names of the model files should be model.ckpt-1000000.data-00000-of-00001, model.ckpt-1000000.index and model.ckpt-1000000.meta.

The error you are getting is because the ckpt files and the python file you are running(run_ner.py) are in different paths.

python run_ner.py \
--do_train=true \
--do_eval=true \
--vocab_file=vocab.txt \
--bert_config_file=bert_config.json \
--init_checkpoint=<path to folder where ckpt files are saved>/model.ckpt-1000000 \
--num_train_epochs=10.0 \
--data_dir=NCBI-disease/ \
--output_dir=epoch1

Don't change --init_checkpoint=model.ckpt-1000000 to --init_checkpoint=model.ckpt-1000000.data-00000-of-00001 while running run_ner.py .

Hope this helps.