Flair: AttributeError when finetuned bert model for downstream task is used

[Michael95-m]

Describe the bug
I got the error called AttributeError:'BertTokenizer' object has no attribute 'model_max_length' when i used model.predict(sentence)

To Reproduce
I didn't get any errors when I trained. The script for training model is

``
from flair.data import Corpus
from flair.datasets import ColumnCorpus
from flair.embeddings import TransformerWordEmbeddings, TokenEmbeddings, StackedEmbeddings, BertEmbeddings, CharacterEmbeddings, BytePairEmbeddings, WordEmbeddings
from flair.visual.training_curves import Plotter
from flair.models import SequenceTagger
from flair.trainers import ModelTrainer
from typing import List
from torch.optim.adam import Adam
import sys
import flair, torch

flair.device = torch.device('cuda:0')
columns = {0: 'text', 1: 'ner'}
data_folder = './data/Model_Trainer_Data'
corpus: Corpus = ColumnCorpus(data_folder, columns,train_file= sys.argv[1] , dev_file=sys.argv[3], test_file= sys.argv[4], in_memory=False)

tag_type = 'ner'
tag_dictionary = corpus.make_tag_dictionary(tag_type=tag_type)

embeddings = TransformerWordEmbeddings(
'bert-base-multilingual-cased', # which transformer model
layers="-1", # which layers (here: only last layer when fine-tuning)
pooling_operation='first_last', # how to pool over split tokens
fine_tune=True, # whether or not to fine-tune
)

tagger: SequenceTagger = SequenceTagger(
hidden_size=256,
embeddings=embeddings,
tag_dictionary=tag_dictionary,
tag_type=tag_type,
use_crf=False,
use_rnn=False,
)

trainer = ModelTrainer(tagger, corpus, optimizer=torch.optim.Adam)

trainer.train('./results/'+ sys.argv[2],
learning_rate=3e-5,
embeddings_storage_mode = 'gpu',
mini_batch_chunk_size=2,
max_epochs=4,
train_with_dev=False,
checkpoint=True)

plotter = Plotter()
plotter.plot_weights('./results/'+ sys.argv[2]+'/weights.txt')
``

The prediction script is below.

Expected behavior
To get no errors in predicting with model

Screenshots

Environment (please complete the following information):

• OS [Linux mint 19.3]:
• Version [flair current lastest branch]:

[alanakbik]

Thanks for reporting this. I can reproduce this error if I train the model on one machine and then predict on another. Is this the same for you? If I train and predict on the same machine, it seems to work.

[alanakbik]

I've merged a PR that fixes this for me. Can you update and try again?

[Michael95-m]

@alanakbik yes bro. I trained and predicted on different machines.

[Michael95-m]

I still got these two errors.

This is another errors.

~/Research/flair/utils_flair/utils/word_break_flair.py in word_break_only(wb_model_path, sent_data)
     70 
     71 def word_break_only(wb_model_path, sent_data):
---> 72     wb_model = SequenceTagger.load(wb_model_path)
     73     if type(sent_data) == str:
     74         wb_str = word_break(sent_data, wb_model)

~/Research/flair/flair/nn.py in load(cls, model)
     84             # see https://github.com/zalandoresearch/flair/issues/351
     85             f = file_utils.load_big_file(str(model_file))
---> 86             state = torch.load(f, map_location=flair.device)
     87 
     88         model = cls._init_model_with_state_dict(state)

~/anaconda3/envs/flair_0.4.5/lib/python3.7/site-packages/torch/serialization.py in load(f, map_location, pickle_module, **pickle_load_args)
    591                     return torch.jit.load(f)
    592                 return _load(opened_zipfile, map_location, pickle_module, **pickle_load_args)
--> 593         return _legacy_load(opened_file, map_location, pickle_module, **pickle_load_args)
    594 
    595 

~/anaconda3/envs/flair_0.4.5/lib/python3.7/site-packages/torch/serialization.py in _legacy_load(f, map_location, pickle_module, **pickle_load_args)
    771     unpickler = pickle_module.Unpickler(f, **pickle_load_args)
    772     unpickler.persistent_load = persistent_load
--> 773     result = unpickler.load()
    774 
    775     deserialized_storage_keys = pickle_module.load(f, **pickle_load_args)

~/Research/flair/flair/embeddings/token.py in __setstate__(self, d)
   1062         # reload tokenizer to get around serialization issues
   1063         model_name = self.name.split('transformer-word-')[-1]
-> 1064         self.tokenizer = AutoTokenizer.from_pretrained(model_name)
   1065 
   1066 

~/anaconda3/envs/flair_0.4.5/lib/python3.7/site-packages/transformers/tokenization_auto.py in from_pretrained(cls, pretrained_model_name_or_path, *inputs, **kwargs)
    187         config = kwargs.pop("config", None)
    188         if not isinstance(config, PretrainedConfig):
--> 189             config = AutoConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
    190 
    191         if "bert-base-japanese" in pretrained_model_name_or_path:

~/anaconda3/envs/flair_0.4.5/lib/python3.7/site-packages/transformers/configuration_auto.py in from_pretrained(cls, pretrained_model_name_or_path, **kwargs)
    190         """
    191         config_dict, _ = PretrainedConfig.get_config_dict(
--> 192             pretrained_model_name_or_path, pretrained_config_archive_map=ALL_PRETRAINED_CONFIG_ARCHIVE_MAP, **kwargs
    193         )
    194 

~/anaconda3/envs/flair_0.4.5/lib/python3.7/site-packages/transformers/configuration_utils.py in get_config_dict(cls, pretrained_model_name_or_path, pretrained_config_archive_map, **kwargs)
    247                 proxies=proxies,
    248                 resume_download=resume_download,
--> 249                 local_files_only=local_files_only,
    250             )
    251             # Load config dict

~/anaconda3/envs/flair_0.4.5/lib/python3.7/site-packages/transformers/file_utils.py in cached_path(url_or_filename, cache_dir, force_download, proxies, resume_download, user_agent, extract_compressed_file, force_extract, local_files_only)
    265             resume_download=resume_download,
    266             user_agent=user_agent,
--> 267             local_files_only=local_files_only,
    268         )
    269     elif os.path.exists(url_or_filename):

~/anaconda3/envs/flair_0.4.5/lib/python3.7/site-packages/transformers/file_utils.py in get_from_cache(url, cache_dir, force_download, proxies, etag_timeout, resume_download, user_agent, local_files_only)
    444             logger.info("%s not found in cache or force_download set to True, downloading to %s", url, temp_file.name)
    445 
--> 446             http_get(url, temp_file, proxies=proxies, resume_size=resume_size, user_agent=user_agent)
    447 
    448         logger.info("storing %s in cache at %s", url, cache_path)

~/anaconda3/envs/flair_0.4.5/lib/python3.7/site-packages/transformers/file_utils.py in http_get(url, temp_file, proxies, resume_size, user_agent)
    335         initial=resume_size,
    336         desc="Downloading",
--> 337         disable=bool(logger.getEffectiveLevel() == logging.NOTSET),
    338     )
    339     for chunk in response.iter_content(chunk_size=1024):

~/anaconda3/envs/flair_0.4.5/lib/python3.7/site-packages/tqdm/notebook.py in __init__(self, *args, **kwargs)
    207         total = self.total * unit_scale if self.total else self.total
    208         self.container = self.status_printer(
--> 209             self.fp, total, self.desc, self.ncols)
    210         self.sp = self.display
    211 

~/anaconda3/envs/flair_0.4.5/lib/python3.7/site-packages/tqdm/notebook.py in status_printer(_, total, desc, ncols)
    102             # #187 #451 #558
    103             raise ImportError(
--> 104                 "FloatProgress not found. Please update jupyter and ipywidgets."
    105                 " See https://ipywidgets.readthedocs.io/en/stable"
    106                 "/user_install.html")

ImportError: FloatProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html

I am not sure this is because of new update but the older model(not transformer model.. I mean bilstm plus crf) works just fine and only transformer model got these errors.

[alanakbik]

Can you share your model to reproduce?

[Michael95-m]

I used the same previous model that got the errors. I didn't retrain it. Should I need to retrain the model?

[alanakbik]

The same model should work.

[Michael95-m]

I think my pytorch version got problem. Which torch version did u use @alanakbik ??

[alanakbik]

I'm using 1.4.0

[Michael95-m]

@alanakbik , it works when running in python script but doesn't work and get errors when running in jupyter notebook. I think my jupyter notebook causes the problems. The updated code works!!! Thank you.