transformers: RuntimeError: unexpected EOF, expected 7491165 more bytes. The file might be corrupted.
❓ Questions & Help
I tried a small chunk of code from the Readme.md
import torch
from transformers import *
MODELS = [(BertModel, BertTokenizer, 'bert-base-uncased')]
for model_class, tokenizer_class, pretrained_weights in MODELS:
# Load pretrained model/tokenizer
tokenizer = tokenizer_class.from_pretrained(pretrained_weights)
model = model_class.from_pretrained(pretrained_weights)
input_ids = torch.tensor([tokenizer.encode("Here is some text to encode", add_special_tokens=True)]) # Add special tokens takes care of adding [CLS], [SEP], <s>... tokens in the right way for each model.
with torch.no_grad():
last_hidden_states = model(input_ids)[0]
It is giving me the following error
RuntimeError Traceback (most recent call last)
<ipython-input-3-6528fe9b0472> in <module>
3 tokenizer = tokenizer_class.from_pretrained(pretrained_weights)
----> 4 model = model_class.from_pretrained(pretrained_weights)
~/.conda/envs/transformers/lib/python3.7/site-packages/transformers/modeling_utils.py in from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs)
343
344 if state_dict is None and not from_tf:
--> 345 state_dict = torch.load(resolved_archive_file, map_location='cpu')
346
347 missing_keys = []
~/.conda/envs/transformers/lib/python3.7/site-packages/torch/serialization.py in load(f, map_location, pickle_module, **pickle_load_args)
424 if sys.version_info >= (3, 0) and 'encoding' not in pickle_load_args.keys():
425 pickle_load_args['encoding'] = 'utf-8'
--> 426 return _load(f, map_location, pickle_module, **pickle_load_args)
427 finally:
428 if new_fd:
~/.conda/envs/transformers/lib/python3.7/site-packages/torch/serialization.py in _load(f, map_location, pickle_module, **pickle_load_args)
618 for key in deserialized_storage_keys:
619 assert key in deserialized_objects
--> 620 deserialized_objects[key]._set_from_file(f, offset, f_should_read_directly)
621 if offset is not None:
622 offset = f.tell()
RuntimeError: unexpected EOF, expected 7491165 more bytes. The file might be corrupted.
Haven’t modified anything in the library.
About this issue
- Original URL
- State: closed
- Created 5 years ago
- Comments: 16 (3 by maintainers)
Hi! It seems to me that the file that was downloaded was corrupted, probably because of lacking space or a network error. Could you try using the
from_pretrainedwith theforce_downloadoption ?If you are using Window 10 machine, deleting
vgg16-somethingin folderC:\Users\UserName\.cache\torch\checkpointswould solve probelm.This worked for me