Unable to download and use any of the models
alisonreboud opened this issue · 3 comments
alisonreboud commented
Hi, I am not able to get the models using the readme snippet
from transformers import AutoTokenizer, T5ForConditionalGeneration
model_name = "allenai/unifiedqa-t5-small" # you can specify the model size here
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = T5ForConditionalGeneration.from_pretrained(model_name)
def run_model(input_string, **generator_args):
input_ids = tokenizer.encode(input_string, return_tensors="pt")
res = model.generate(input_ids, **generator_args)
return tokenizer.batch_decode(res, `skip_special_tokens=True)
I get the following error
---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
/usr/lib/python3.6/tarfile.py in nti(s)
188 s = nts(s, "ascii", "strict")
--> 189 n = int(s.strip() or "0", 8)
190 except ValueError:
ValueError: invalid literal for int() with base 8: 'v2\nq\x03((X'
During handling of the above exception, another exception occurred:
InvalidHeaderError Traceback (most recent call last)
/usr/lib/python3.6/tarfile.py in next(self)
2298 try:
-> 2299 tarinfo = self.tarinfo.fromtarfile(self)
2300 except EOFHeaderError as e:
/usr/lib/python3.6/tarfile.py in fromtarfile(cls, tarfile)
1092 buf = tarfile.fileobj.read(BLOCKSIZE)
-> 1093 obj = cls.frombuf(buf, tarfile.encoding, tarfile.errors)
1094 obj.offset = tarfile.fileobj.tell() - BLOCKSIZE
/usr/lib/python3.6/tarfile.py in frombuf(cls, buf, encoding, errors)
1034
-> 1035 chksum = nti(buf[148:156])
1036 if chksum not in calc_chksums(buf):
/usr/lib/python3.6/tarfile.py in nti(s)
190 except ValueError:
--> 191 raise InvalidHeaderError("invalid header")
192 return n
InvalidHeaderError: invalid header
During handling of the above exception, another exception occurred:
ReadError Traceback (most recent call last)
/usr/local/lib/python3.6/dist-packages/torch/serialization.py in _load(f, map_location, pickle_module, **pickle_load_args)
554 >>> torch.load('tensors.pt', map_location=lambda storage, loc: storage)
--> 555 # Load all tensors onto GPU 1
556 >>> torch.load('tensors.pt', map_location=lambda storage, loc: storage.cuda(1))
/usr/local/lib/python3.6/dist-packages/torch/serialization.py in legacy_load(f)
465 pickler.persistent_id = persistent_id
--> 466 pickler.dump(obj)
467 data_value = data_buf.getvalue()
/usr/lib/python3.6/tarfile.py in open(cls, name, mode, fileobj, bufsize, **kwargs)
1590 raise CompressionError("unknown compression type %r" % comptype)
-> 1591 return func(name, filemode, fileobj, **kwargs)
1592
/usr/lib/python3.6/tarfile.py in taropen(cls, name, mode, fileobj, **kwargs)
1620 raise ValueError("mode must be 'r', 'a', 'w' or 'x'")
-> 1621 return cls(name, mode, fileobj, **kwargs)
1622
/usr/lib/python3.6/tarfile.py in __init__(self, name, mode, fileobj, format, tarinfo, dereference, ignore_zeros, encoding, errors, pax_headers, debug, errorlevel, copybufsize)
1483 self.firstmember = None
-> 1484 self.firstmember = self.next()
1485
/usr/lib/python3.6/tarfile.py in next(self)
2310 elif self.offset == 0:
-> 2311 raise ReadError(str(e))
2312 except EmptyHeaderError:
ReadError: invalid header
During handling of the above exception, another exception occurred:
RuntimeError Traceback (most recent call last)
/usr/local/lib/python3.6/dist-packages/transformers/modeling_utils.py in from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs)
855 from_tf = kwargs.pop("from_tf", False)
--> 856 force_download = kwargs.pop("force_download", False)
857 resume_download = kwargs.pop("resume_download", False)
/usr/local/lib/python3.6/dist-packages/torch/serialization.py in load(f, map_location, pickle_module, **pickle_load_args)
385 try:
--> 386 source_lines, _, source_file = get_source_lines_and_file(obj)
387 source = ''.join(source_lines)
/usr/local/lib/python3.6/dist-packages/torch/serialization.py in _load(f, map_location, pickle_module, **pickle_load_args)
558 >>> torch.load('tensors.pt', map_location={'cuda:1':'cuda:0'})
--> 559 # Load tensor from io.BytesIO object
560 >>> with open('tensor.pt', 'rb') as f:
RuntimeError: /root/.cache/torch/transformers/1fb349953ea9c104fd662676908c1c8496dfe57a33cb2e23b091055f49f76295.1f377393c926073ea459ce25ddbef3e4f3e9be2e4f3bf2d7684d211ff9740b52 is a zip archive (did you mean to use torch.jit.load()?)
During handling of the above exception, another exception occurred:
OSError Traceback (most recent call last)
<ipython-input-51-72f3a3fc32af> in <module>
3 model_name = "allenai/unifiedqa-t5-small" # you can specify the model size here
4 tokenizer = AutoTokenizer.from_pretrained(model_name)
----> 5 model = T5ForConditionalGeneration.from_pretrained(model_name)
6
7 def run_model(input_string, **generator_args):
/usr/local/lib/python3.6/dist-packages/transformers/modeling_utils.py in from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs)
857 resume_download = kwargs.pop("resume_download", False)
858 proxies = kwargs.pop("proxies", None)
--> 859 output_loading_info = kwargs.pop("output_loading_info", False)
860 local_files_only = kwargs.pop("local_files_only", False)
861 revision = kwargs.pop("revision", None)
OSError: Unable to load weights from pytorch checkpoint file. If you tried to load a PyTorch model from a TF 2.0 checkpoint, please set from_tf=True. "
Trying with from_tf=True, I get
`OSError Traceback (most recent call last)
/usr/local/lib/python3.6/dist-packages/transformers/modeling_utils.py in from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs)
834 ``kwargs`` that corresponds to a configuration attribute will be used to override said attribute
--> 835 with the supplied ``kwargs`` value. Remaining keys that do not correspond to any configuration
836 attribute will be passed to the underlying model's ``__init__`` function.
OSError:
During handling of the above exception, another exception occurred:
OSError Traceback (most recent call last)
<ipython-input-52-ad634a93ae8f> in <module>
3 model_name = "allenai/unifiedqa-t5-small" # you can specify the model size here
4 tokenizer = AutoTokenizer.from_pretrained(model_name)
----> 5 model = T5ForConditionalGeneration.from_pretrained(model_name,from_tf=True)
6
7 def run_model(input_string, **generator_args):
/usr/local/lib/python3.6/dist-packages/transformers/modeling_utils.py in from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs)
840 >>> from transformers import BertConfig, BertModel
841 >>> # Download model and configuration from S3 and cache.
--> 842 >>> model = BertModel.from_pretrained('bert-base-uncased')
843 >>> # Model was saved using `save_pretrained('./test/saved_model/')` (for example purposes, not runnable).
844 >>> model = BertModel.from_pretrained('./test/saved_model/')
OSError: Can't load weights for 'allenai/unifiedqa-t5-small'. Make sure that:
- 'allenai/unifiedqa-t5-small' is a correct model identifier listed on 'https://huggingface.co/models'
- or 'allenai/unifiedqa-t5-small' is the correct path to a directory containing a file named one of pytorch_model.bin, tf_model.h5, model.ckpt.`
-
Thanks a lot for your help!
danyaljj commented
Thanks for reporting the issue!
This is very likely due to the incompatibility of your transformers and pytorch. See this:
huggingface/transformers#7594
alisonreboud commented
Hello thanks for your answer,
actually I was able to solve it upgrading my python version and running this
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
tokenizer = AutoTokenizer.from_pretrained("allenai/unifiedqa-t5-small")
model = AutoModelForSeq2SeqLM.from_pretrained("allenai/unifiedqa-t5-small")
I had another question : is it possible to get a confidence score as ouput ( I run the model on different text parts and only one has the answer so I would like to keep the one with the highest confidence). I have seen that it should be possible to add a output_scores arg in generate() but it doesnt seem to work)