Issue regarding "9.1 Download pretrained transformers automatically"

[CaterinaFabbri]

While running def translate_a_single_sentence(translation_config): I have encountered an error in which the file en-de.tgz is not recognized as a gzip file. How could I do? Below, it is reported the snippet of the error : ` downloading en-de.tgz C:\Users..\pytorch-original-transformer\data\iwslt\en-de.tgz: 97.4kB [00:00, 1.60MB/s]

BadGzipFile Traceback (most recent call last) ~\anaconda3\envs\pytorch-transformer\lib\tarfile.py in gzopen(cls, name, mode, fileobj, compresslevel, kwargs) 1669 try: -> 1670 t = cls.taropen(name, mode, fileobj, kwargs) 1671 except OSError:

~\anaconda3\envs\pytorch-transformer\lib\tarfile.py in taropen(cls, name, mode, fileobj, kwargs) 1646 raise ValueError("mode must be 'r', 'a', 'w' or 'x'") -> 1647 return cls(name, mode, fileobj, kwargs) 1648

~\anaconda3\envs\pytorch-transformer\lib\tarfile.py in init(self, name, mode, fileobj, format, tarinfo, dereference, ignore_zeros, encoding, errors, pax_headers, debug, errorlevel, copybufsize) 1509 self.firstmember = None -> 1510 self.firstmember = self.next() 1511

~\anaconda3\envs\pytorch-transformer\lib\tarfile.py in next(self) 2310 try: -> 2311 tarinfo = self.tarinfo.fromtarfile(self) 2312 except EOFHeaderError as e:

~\anaconda3\envs\pytorch-transformer\lib\tarfile.py in fromtarfile(cls, tarfile) 1101 """ -> 1102 buf = tarfile.fileobj.read(BLOCKSIZE) 1103 obj = cls.frombuf(buf, tarfile.encoding, tarfile.errors)

~\anaconda3\envs\pytorch-transformer\lib\gzip.py in read(self, size) 291 raise OSError(errno.EBADF, "read() on write-only GzipFile object") --> 292 return self._buffer.read(size) 293

~\anaconda3\envs\pytorch-transformer\lib_compression.py in readinto(self, b) 67 with memoryview(b) as view, view.cast("B") as byte_view: ---> 68 data = self.read(len(byte_view)) 69 byte_view[:len(data)] = data

~\anaconda3\envs\pytorch-transformer\lib\gzip.py in read(self, size) 478 self._init_read() --> 479 if not self._read_gzip_header(): 480 self._size = self._pos

~\anaconda3\envs\pytorch-transformer\lib\gzip.py in _read_gzip_header(self) 426 if magic != b'\037\213': --> 427 raise BadGzipFile('Not a gzipped file (%r)' % magic) 428

BadGzipFile: Not a gzipped file (b'<!')

During handling of the above exception, another exception occurred:

ReadError Traceback (most recent call last)

in 85 86 # Translate the given source sentence ---> 87 translate_a_single_sentence(translation_config) in translate_a_single_sentence(translation_config) 5 print(2) 6 # Step 1: Prepare the field processor (tokenizer, numericalizer) ----> 7 _, _, src_field_processor, trg_field_processor = get_datasets_and_vocabs( 8 translation_config['dataset_path'], 9 translation_config['language_direction'], in get_datasets_and_vocabs(dataset_path, language_direction, use_iwslt, use_caching_mechanism) 41 dataset_split_fn = datasets.IWSLT.splits if use_iwslt else datasets.WMT14.splits 42 ---> 43 train_dataset, val_dataset, test_dataset = dataset_split_fn( 44 exts=(src_ext, trg_ext), 45 fields=fields, ~\anaconda3\envs\pytorch-transformer\lib\site-packages\torchtext\datasets\translation.py in splits(cls, exts, fields, root, train, validation, test, **kwargs) 142 cls.urls = [cls.base_url.format(exts[0][1:], exts[1][1:], cls.dirname)] 143 check = os.path.join(root, cls.name, cls.dirname) --> 144 path = cls.download(root, check=check) 145 146 train = '.'.join([train, cls.dirname]) ~\anaconda3\envs\pytorch-transformer\lib\site-packages\torchtext\data\dataset.py in download(cls, root, check) 189 # tarfile cannot handle bare .gz files 190 elif ext == '.tgz' or ext == '.gz' and ext_inner == '.tar': --> 191 with tarfile.open(zpath, 'r:gz') as tar: 192 dirs = [member for member in tar.getmembers()] 193 tar.extractall(path=path, members=dirs) ~\anaconda3\envs\pytorch-transformer\lib\tarfile.py in open(cls, name, mode, fileobj, bufsize, **kwargs) 1615 else: 1616 raise CompressionError("unknown compression type %r" % comptype) -> 1617 return func(name, filemode, fileobj, **kwargs) 1618 1619 elif "|" in mode: ~\anaconda3\envs\pytorch-transformer\lib\tarfile.py in gzopen(cls, name, mode, fileobj, compresslevel, **kwargs) 1672 fileobj.close() 1673 if mode == 'r': -> 1674 raise ReadError("not a gzip file") 1675 raise 1676 except: ReadError: not a gzip file` Thank you very much!

[CaterinaFabbri]

The issue regards the dataset. I replaced it with Multi30k. from torchtext.datasets import Multi30k ...... train_dataset, val_dataset, test_dataset = Multi30k.splits(exts=(src_ext, trg_ext), fields=fields,)

[hsuBnOediH]

The issue regards the dataset. I replaced it with Multi30k. from torchtext.datasets import Multi30k ...... train_dataset, val_dataset, test_dataset = Multi30k.splits(exts=(src_ext, trg_ext), fields=fields,)

Thank you SO MUCH for sharing the solution, I sucked on the url for a whole afternoon! 感谢这个B