Datasets:

allenai

/

dolma

like 856

Follow

Ai2 1.76k

Hi!
I'm trying to use the dataset in streaming mode. When I try to download the first instance, I get the following error:

ValueError: The HTTP server doesn't appear to support range requests. Only reading this file from the beginning is supported. Open with block_size=0 for a streaming file interface.

Here's the code returning the error:

dataset = load_dataset("allenai/dolma", "v1_6", streaming=True, split='train')
ds_iterator = iter(dataset)
data_sample = next(ds_iterator)

Here's the whole error stack:

/usr/local/lib/python3.10/dist-packages/datasets/iterable_dataset.py in __iter__(self)
   1386             return
   1387 
-> 1388         for key, example in ex_iterable:
   1389             if self.features:
   1390                 # `IterableDataset` automatically fills missing columns with None.

/usr/local/lib/python3.10/dist-packages/datasets/iterable_dataset.py in __iter__(self)
    985         # this is the shuffle buffer that we keep in memory
    986         mem_buffer = []
--> 987         for x in self.ex_iterable:
    988             if len(mem_buffer) == buffer_size:  # if the buffer is full, pick and example from it
    989                 i = next(indices_iterator)

/usr/local/lib/python3.10/dist-packages/datasets/iterable_dataset.py in __iter__(self)
    260         rng = deepcopy(self.generator)
    261         kwargs_with_shuffled_shards = _shuffle_gen_kwargs(rng, self.kwargs)
--> 262         yield from self.generate_examples_fn(**kwargs_with_shuffled_shards)
    263 
    264     def shard_data_sources(self, worker_id: int, num_workers: int) -> "ExamplesIterable":

~/.cache/huggingface/modules/datasets_modules/datasets/allenai--dolma/4ffcc06d84b368ddbd84d7426b091d2e9242e951fb741da8a4f78d4d638d2ea4/dolma.py in _generate_examples(self, files)
    127 
    128             with gzip.open(fn, mode="rt", encoding="utf-8") as f:
--> 129                 for line in f:
    130                     row = json.loads(line)
    131                     yield row["id"], {

/usr/lib/python3.10/gzip.py in read1(self, size)
    312         if size < 0:
    313             size = io.DEFAULT_BUFFER_SIZE
--> 314         return self._buffer.read1(size)
    315 
    316     def peek(self, n):

/usr/lib/python3.10/_compression.py in readinto(self, b)
     66     def 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
     70         return len(data)

/usr/lib/python3.10/gzip.py in read(self, size)
    492 
    493             # Read a chunk of data from the file
--> 494             buf = self._fp.read(io.DEFAULT_BUFFER_SIZE)
    495 
    496             uncompress = self._decompressor.decompress(buf, size)

/usr/lib/python3.10/gzip.py in read(self, size)
     95             self._read = None
     96             return self._buffer[read:] + \
---> 97                    self.file.read(size-self._length+read)
     98 
     99     def prepend(self, prepend=b''):

/usr/local/lib/python3.10/dist-packages/datasets/download/streaming_download_manager.py in read_with_retries(*args, **kwargs)
    340         for retry in range(1, max_retries + 1):
    341             try:
--> 342                 out = read(*args, **kwargs)
    343                 break
    344             except (ClientError, TimeoutError) as err:

/usr/local/lib/python3.10/dist-packages/fsspec/implementations/http.py in read(self, length)
    598         else:
    599             length = min(self.size - self.loc, length)
--> 600         return super().read(length)
    601 
    602     async def async_fetch_all(self):

/usr/local/lib/python3.10/dist-packages/fsspec/spec.py in read(self, length)
   1788             # don't even bother calling fetch
   1789             return b""
-> 1790         out = self.cache._fetch(self.loc, self.loc + length)
   1791         self.loc += len(out)
   1792         return out

/usr/local/lib/python3.10/dist-packages/fsspec/caching.py in _fetch(self, start, end)
    394                 self.start = start
    395             else:
--> 396                 new = self.fetcher(self.end, bend)
    397                 self.cache = self.cache + new
    398 

/usr/local/lib/python3.10/dist-packages/fsspec/asyn.py in wrapper(*args, **kwargs)
    119     def wrapper(*args, **kwargs):
    120         self = obj or args[0]
--> 121         return sync(self.loop, func, *args, **kwargs)
    122 
    123     return wrapper

/usr/local/lib/python3.10/dist-packages/fsspec/asyn.py in sync(loop, func, timeout, *args, **kwargs)
    104         raise FSTimeoutError from return_result
    105     elif isinstance(return_result, BaseException):
--> 106         raise return_result
    107     else:
    108         return return_result

/usr/local/lib/python3.10/dist-packages/fsspec/asyn.py in _runner(event, coro, result, timeout)
     59         coro = asyncio.wait_for(coro, timeout=timeout)
     60     try:
---> 61         result[0] = await coro
     62     except Exception as ex:
     63         result[0] = ex

/usr/local/lib/python3.10/dist-packages/fsspec/implementations/http.py in async_fetch_range(self, start, end)
    669                 out = await r.read()
    670             elif start > 0:
--> 671                 raise ValueError(
    672                     "The HTTP server doesn't appear to support range requests. "
    673                     "Only reading this file from the beginning is supported. "

ValueError: The HTTP server doesn't appear to support range requests. Only reading this file from the beginning is supported. Open with block_size=0 for a streaming file interface.

Can I do something to prevent this error, or is it a configuration error on the maintainer's side?
Thank you in advance!