Fix conda release by adding pyarrow-hotfix dependency

[albertvillanova]

Fix conda release by adding pyarrow-hotfix dependency.

Note that conda release failed in latest 2.14.7 release: https://github.com/huggingface/datasets/actions/runs/6874667214/job/18696761723

Traceback (most recent call last):
  File "/usr/share/miniconda/envs/build-datasets/conda-bld/datasets_1700036460222/test_tmp/run_test.py", line 2, in <module>
    import datasets
  File "/usr/share/miniconda/envs/build-datasets/conda-bld/datasets_1700036460222/_test_env_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold/lib/python3.12/site-packages/datasets/__init__.py", line 22, in <module>
    from .arrow_dataset import Dataset
  File "/usr/share/miniconda/envs/build-datasets/conda-bld/datasets_1700036460222/_test_env_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold/lib/python3.12/site-packages/datasets/arrow_dataset.py", line 67, in <module>
    from .arrow_writer import ArrowWriter, OptimizedTypedSequence
  File "/usr/share/miniconda/envs/build-datasets/conda-bld/datasets_1700036460222/_test_env_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold/lib/python3.12/site-packages/datasets/arrow_writer.py", line 27, in <module>
    from .features import Features, Image, Value
  File "/usr/share/miniconda/envs/build-datasets/conda-bld/datasets_1700036460222/_test_env_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold/lib/python3.12/site-packages/datasets/features/__init__.py", line 18, in <module>
    from .features import Array2D, Array3D, Array4D, Array5D, ClassLabel, Features, Sequence, Value
  File "/usr/share/miniconda/envs/build-datasets/conda-bld/datasets_1700036460222/_test_env_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold/lib/python3.12/site-packages/datasets/features/features.py", line 34, in <module>
    import pyarrow_hotfix  # noqa: F401  # to fix vulnerability on pyarrow<14.0.1
    ^^^^^^^^^^^^^^^^^^^^^
ModuleNotFoundError: No module named 'pyarrow_hotfix'

[HuggingFaceDocBuilderDev]

The documentation is not available anymore as the PR was closed or merged.

[github-actions]

Show benchmarks

PyArrow==8.0.0

Show updated benchmarks!

Benchmark: benchmark_array_xd.json

metric	read_batch_formatted_as_numpy after write_array2d	read_batch_formatted_as_numpy after write_flattened_sequence	read_batch_formatted_as_numpy after write_nested_sequence	read_batch_unformated after write_array2d	read_batch_unformated after write_flattened_sequence	read_batch_unformated after write_nested_sequence	read_col_formatted_as_numpy after write_array2d	read_col_formatted_as_numpy after write_flattened_sequence	read_col_formatted_as_numpy after write_nested_sequence	read_col_unformated after write_array2d	read_col_unformated after write_flattened_sequence	read_col_unformated after write_nested_sequence	read_formatted_as_numpy after write_array2d	read_formatted_as_numpy after write_flattened_sequence	read_formatted_as_numpy after write_nested_sequence	read_unformated after write_array2d	read_unformated after write_flattened_sequence	read_unformated after write_nested_sequence	write_array2d	write_flattened_sequence	write_nested_sequence
new / old (diff)	0.004476 / 0.011353 (-0.006877)	0.002691 / 0.011008 (-0.008317)	0.061400 / 0.038508 (0.022892)	0.030096 / 0.023109 (0.006986)	0.279868 / 0.275898 (0.003970)	0.310320 / 0.323480 (-0.013159)	0.003873 / 0.007986 (-0.004112)	0.002394 / 0.004328 (-0.001935)	0.048307 / 0.004250 (0.044056)	0.043326 / 0.037052 (0.006273)	0.288256 / 0.258489 (0.029767)	0.311449 / 0.293841 (0.017609)	0.022970 / 0.128546 (-0.105576)	0.006714 / 0.075646 (-0.068932)	0.201656 / 0.419271 (-0.217615)	0.052811 / 0.043533 (0.009278)	0.285123 / 0.255139 (0.029984)	0.301495 / 0.283200 (0.018295)	0.017531 / 0.141683 (-0.124152)	1.097660 / 1.452155 (-0.354494)	1.161986 / 1.492716 (-0.330731)

Benchmark: benchmark_getitem_100B.json

metric	get_batch_of_1024_random_rows	get_batch_of_1024_rows	get_first_row	get_last_row
new / old (diff)	0.089223 / 0.018006 (0.071217)	0.297815 / 0.000490 (0.297326)	0.000205 / 0.000200 (0.000005)	0.000042 / 0.000054 (-0.000013)

Benchmark: benchmark_indices_mapping.json

metric	select	shard	shuffle	sort	train_test_split
new / old (diff)	0.018679 / 0.037411 (-0.018732)	0.062742 / 0.014526 (0.048216)	0.072869 / 0.176557 (-0.103687)	0.120730 / 0.737135 (-0.616406)	0.074526 / 0.296338 (-0.221813)

Benchmark: benchmark_iterating.json

metric	read 5000	read 50000	read_batch 50000 10	read_batch 50000 100	read_batch 50000 1000	read_formatted numpy 5000	read_formatted pandas 5000	read_formatted tensorflow 5000	read_formatted torch 5000	read_formatted_batch numpy 5000 10	read_formatted_batch numpy 5000 1000	shuffled read 5000	shuffled read 50000	shuffled read_batch 50000 10	shuffled read_batch 50000 100	shuffled read_batch 50000 1000	shuffled read_formatted numpy 5000	shuffled read_formatted_batch numpy 5000 10	shuffled read_formatted_batch numpy 5000 1000
new / old (diff)	0.299977 / 0.215209 (0.084768)	2.921029 / 2.077655 (0.843375)	1.632283 / 1.504120 (0.128163)	1.508008 / 1.541195 (-0.033187)	1.513967 / 1.468490 (0.045477)	0.403056 / 4.584777 (-4.181721)	2.340011 / 3.745712 (-1.405701)	2.552319 / 5.269862 (-2.717543)	1.549741 / 4.565676 (-3.015935)	0.046303 / 0.424275 (-0.377972)	0.004768 / 0.007607 (-0.002839)	0.356921 / 0.226044 (0.130877)	3.506410 / 2.268929 (1.237482)	1.975394 / 55.444624 (-53.469230)	1.688683 / 6.876477 (-5.187794)	1.715502 / 2.142072 (-0.426571)	0.471016 / 4.805227 (-4.334212)	0.099552 / 6.500664 (-6.401112)	0.042095 / 0.075469 (-0.033374)

Benchmark: benchmark_map_filter.json

metric	filter	map fast-tokenizer batched	map identity	map identity batched	map no-op batched	map no-op batched numpy	map no-op batched pandas	map no-op batched pytorch	map no-op batched tensorflow
new / old (diff)	0.955784 / 1.841788 (-0.886004)	11.191802 / 8.074308 (3.117494)	10.127818 / 10.191392 (-0.063574)	0.141225 / 0.680424 (-0.539199)	0.014486 / 0.534201 (-0.519715)	0.267204 / 0.579283 (-0.312079)	0.289108 / 0.434364 (-0.145256)	0.309458 / 0.540337 (-0.230880)	0.422802 / 1.386936 (-0.964134)

PyArrow==latest

Show updated benchmarks!

Benchmark: benchmark_array_xd.json

metric	read_batch_formatted_as_numpy after write_array2d	read_batch_formatted_as_numpy after write_flattened_sequence	read_batch_formatted_as_numpy after write_nested_sequence	read_batch_unformated after write_array2d	read_batch_unformated after write_flattened_sequence	read_batch_unformated after write_nested_sequence	read_col_formatted_as_numpy after write_array2d	read_col_formatted_as_numpy after write_flattened_sequence	read_col_formatted_as_numpy after write_nested_sequence	read_col_unformated after write_array2d	read_col_unformated after write_flattened_sequence	read_col_unformated after write_nested_sequence	read_formatted_as_numpy after write_array2d	read_formatted_as_numpy after write_flattened_sequence	read_formatted_as_numpy after write_nested_sequence	read_unformated after write_array2d	read_unformated after write_flattened_sequence	read_unformated after write_nested_sequence	write_array2d	write_flattened_sequence	write_nested_sequence
new / old (diff)	0.004797 / 0.011353 (-0.006556)	0.002907 / 0.011008 (-0.008101)	0.047666 / 0.038508 (0.009158)	0.051183 / 0.023109 (0.028074)	0.266315 / 0.275898 (-0.009583)	0.286429 / 0.323480 (-0.037051)	0.003954 / 0.007986 (-0.004031)	0.002041 / 0.004328 (-0.002288)	0.047652 / 0.004250 (0.043401)	0.038211 / 0.037052 (0.001158)	0.272210 / 0.258489 (0.013721)	0.299425 / 0.293841 (0.005584)	0.024266 / 0.128546 (-0.104280)	0.006747 / 0.075646 (-0.068900)	0.052959 / 0.419271 (-0.366312)	0.032094 / 0.043533 (-0.011439)	0.265677 / 0.255139 (0.010538)	0.285373 / 0.283200 (0.002174)	0.017577 / 0.141683 (-0.124106)	1.114514 / 1.452155 (-0.337640)	1.212970 / 1.492716 (-0.279746)

Benchmark: benchmark_getitem_100B.json

metric	get_batch_of_1024_random_rows	get_batch_of_1024_rows	get_first_row	get_last_row
new / old (diff)	0.088347 / 0.018006 (0.070341)	0.296678 / 0.000490 (0.296188)	0.000209 / 0.000200 (0.000009)	0.000044 / 0.000054 (-0.000010)

Benchmark: benchmark_indices_mapping.json

metric	select	shard	shuffle	sort	train_test_split
new / old (diff)	0.021159 / 0.037411 (-0.016253)	0.069886 / 0.014526 (0.055360)	0.079832 / 0.176557 (-0.096725)	0.115512 / 0.737135 (-0.621623)	0.081600 / 0.296338 (-0.214739)

Benchmark: benchmark_iterating.json

metric	read 5000	read 50000	read_batch 50000 10	read_batch 50000 100	read_batch 50000 1000	read_formatted numpy 5000	read_formatted pandas 5000	read_formatted tensorflow 5000	read_formatted torch 5000	read_formatted_batch numpy 5000 10	read_formatted_batch numpy 5000 1000	shuffled read 5000	shuffled read 50000	shuffled read_batch 50000 10	shuffled read_batch 50000 100	shuffled read_batch 50000 1000	shuffled read_formatted numpy 5000	shuffled read_formatted_batch numpy 5000 10	shuffled read_formatted_batch numpy 5000 1000
new / old (diff)	0.292659 / 0.215209 (0.077450)	2.872556 / 2.077655 (0.794901)	1.573017 / 1.504120 (0.068897)	1.445122 / 1.541195 (-0.096072)	1.485584 / 1.468490 (0.017094)	0.388638 / 4.584777 (-4.196139)	2.434847 / 3.745712 (-1.310865)	2.518167 / 5.269862 (-2.751695)	1.503000 / 4.565676 (-3.062676)	0.045123 / 0.424275 (-0.379153)	0.004778 / 0.007607 (-0.002829)	0.347955 / 0.226044 (0.121910)	3.384819 / 2.268929 (1.115891)	1.920185 / 55.444624 (-53.524439)	1.646910 / 6.876477 (-5.229567)	1.638092 / 2.142072 (-0.503980)	0.450535 / 4.805227 (-4.354692)	0.095301 / 6.500664 (-6.405363)	0.040275 / 0.075469 (-0.035194)

Benchmark: benchmark_map_filter.json

metric	filter	map fast-tokenizer batched	map identity	map identity batched	map no-op batched	map no-op batched numpy	map no-op batched pandas	map no-op batched pytorch	map no-op batched tensorflow
new / old (diff)	0.956088 / 1.841788 (-0.885700)	11.776642 / 8.074308 (3.702334)	10.651063 / 10.191392 (0.459671)	0.127079 / 0.680424 (-0.553345)	0.015080 / 0.534201 (-0.519121)	0.273737 / 0.579283 (-0.305546)	0.271434 / 0.434364 (-0.162929)	0.308448 / 0.540337 (-0.231889)	0.412467 / 1.386936 (-0.974469)

[albertvillanova]

Once this PR is merged, we should upload the missing version to conda.

@lhoestq you did this in the past. If you tell me your approach (I see a tag called VERSION...), I could do it myself.

[lhoestq]

Maybe open a PR against the 2.14 branch and update release-conda.yml like this ?

- on:
-   push:
-     tags:
-       - "[0-9]+.[0-9]+.[0-9]+*"
+ on: push

and then set it back to normal after the release is done

[albertvillanova]

After having cherry-picked the commit in this PR, I have released the conda package. See:

• https://github.com/huggingface/datasets/actions/runs/6880182419/job/18713812449
• https://anaconda.org/HuggingFace/datasets/files?version=2.14.7

I am merging this PR.

[github-actions]

Show benchmarks

PyArrow==8.0.0

Show updated benchmarks!

Benchmark: benchmark_array_xd.json

metric	read_batch_formatted_as_numpy after write_array2d	read_batch_formatted_as_numpy after write_flattened_sequence	read_batch_formatted_as_numpy after write_nested_sequence	read_batch_unformated after write_array2d	read_batch_unformated after write_flattened_sequence	read_batch_unformated after write_nested_sequence	read_col_formatted_as_numpy after write_array2d	read_col_formatted_as_numpy after write_flattened_sequence	read_col_formatted_as_numpy after write_nested_sequence	read_col_unformated after write_array2d	read_col_unformated after write_flattened_sequence	read_col_unformated after write_nested_sequence	read_formatted_as_numpy after write_array2d	read_formatted_as_numpy after write_flattened_sequence	read_formatted_as_numpy after write_nested_sequence	read_unformated after write_array2d	read_unformated after write_flattened_sequence	read_unformated after write_nested_sequence	write_array2d	write_flattened_sequence	write_nested_sequence
new / old (diff)	0.004993 / 0.011353 (-0.006360)	0.002964 / 0.011008 (-0.008044)	0.062588 / 0.038508 (0.024080)	0.030794 / 0.023109 (0.007685)	0.234856 / 0.275898 (-0.041042)	0.264807 / 0.323480 (-0.058673)	0.003139 / 0.007986 (-0.004847)	0.002498 / 0.004328 (-0.001831)	0.048058 / 0.004250 (0.043807)	0.048349 / 0.037052 (0.011296)	0.238210 / 0.258489 (-0.020279)	0.278144 / 0.293841 (-0.015697)	0.023219 / 0.128546 (-0.105327)	0.007296 / 0.075646 (-0.068351)	0.203263 / 0.419271 (-0.216008)	0.058844 / 0.043533 (0.015311)	0.246330 / 0.255139 (-0.008809)	0.264550 / 0.283200 (-0.018649)	0.018580 / 0.141683 (-0.123103)	1.084163 / 1.452155 (-0.367992)	1.154891 / 1.492716 (-0.337825)

Benchmark: benchmark_getitem_100B.json

metric	get_batch_of_1024_random_rows	get_batch_of_1024_rows	get_first_row	get_last_row
new / old (diff)	0.092393 / 0.018006 (0.074387)	0.300545 / 0.000490 (0.300055)	0.000203 / 0.000200 (0.000003)	0.000047 / 0.000054 (-0.000008)

Benchmark: benchmark_indices_mapping.json

metric	select	shard	shuffle	sort	train_test_split
new / old (diff)	0.018648 / 0.037411 (-0.018763)	0.063151 / 0.014526 (0.048625)	0.074206 / 0.176557 (-0.102350)	0.120929 / 0.737135 (-0.616207)	0.075970 / 0.296338 (-0.220368)

Benchmark: benchmark_iterating.json

metric	read 5000	read 50000	read_batch 50000 10	read_batch 50000 100	read_batch 50000 1000	read_formatted numpy 5000	read_formatted pandas 5000	read_formatted tensorflow 5000	read_formatted torch 5000	read_formatted_batch numpy 5000 10	read_formatted_batch numpy 5000 1000	shuffled read 5000	shuffled read 50000	shuffled read_batch 50000 10	shuffled read_batch 50000 100	shuffled read_batch 50000 1000	shuffled read_formatted numpy 5000	shuffled read_formatted_batch numpy 5000 10	shuffled read_formatted_batch numpy 5000 1000
new / old (diff)	0.278489 / 0.215209 (0.063279)	2.664804 / 2.077655 (0.587150)	1.433040 / 1.504120 (-0.071080)	1.321416 / 1.541195 (-0.219779)	1.320964 / 1.468490 (-0.147526)	0.401289 / 4.584777 (-4.183488)	2.365310 / 3.745712 (-1.380402)	2.635798 / 5.269862 (-2.634063)	1.584384 / 4.565676 (-2.981293)	0.045675 / 0.424275 (-0.378600)	0.004854 / 0.007607 (-0.002753)	0.337592 / 0.226044 (0.111548)	3.330462 / 2.268929 (1.061534)	1.794507 / 55.444624 (-53.650117)	1.531284 / 6.876477 (-5.345193)	1.507165 / 2.142072 (-0.634908)	0.478622 / 4.805227 (-4.326606)	0.099105 / 6.500664 (-6.401560)	0.041575 / 0.075469 (-0.033894)

Benchmark: benchmark_map_filter.json

metric	filter	map fast-tokenizer batched	map identity	map identity batched	map no-op batched	map no-op batched numpy	map no-op batched pandas	map no-op batched pytorch	map no-op batched tensorflow
new / old (diff)	0.941790 / 1.841788 (-0.899997)	11.609871 / 8.074308 (3.535563)	10.770869 / 10.191392 (0.579477)	0.138931 / 0.680424 (-0.541493)	0.014406 / 0.534201 (-0.519795)	0.269681 / 0.579283 (-0.309602)	0.260556 / 0.434364 (-0.173808)	0.308244 / 0.540337 (-0.232093)	0.428867 / 1.386936 (-0.958069)

PyArrow==latest

Show updated benchmarks!

Benchmark: benchmark_array_xd.json

metric	read_batch_formatted_as_numpy after write_array2d	read_batch_formatted_as_numpy after write_flattened_sequence	read_batch_formatted_as_numpy after write_nested_sequence	read_batch_unformated after write_array2d	read_batch_unformated after write_flattened_sequence	read_batch_unformated after write_nested_sequence	read_col_formatted_as_numpy after write_array2d	read_col_formatted_as_numpy after write_flattened_sequence	read_col_formatted_as_numpy after write_nested_sequence	read_col_unformated after write_array2d	read_col_unformated after write_flattened_sequence	read_col_unformated after write_nested_sequence	read_formatted_as_numpy after write_array2d	read_formatted_as_numpy after write_flattened_sequence	read_formatted_as_numpy after write_nested_sequence	read_unformated after write_array2d	read_unformated after write_flattened_sequence	read_unformated after write_nested_sequence	write_array2d	write_flattened_sequence	write_nested_sequence
new / old (diff)	0.004803 / 0.011353 (-0.006550)	0.003263 / 0.011008 (-0.007745)	0.049143 / 0.038508 (0.010635)	0.052033 / 0.023109 (0.028924)	0.267815 / 0.275898 (-0.008083)	0.288733 / 0.323480 (-0.034747)	0.004159 / 0.007986 (-0.003826)	0.002407 / 0.004328 (-0.001921)	0.048978 / 0.004250 (0.044728)	0.038994 / 0.037052 (0.001942)	0.264028 / 0.258489 (0.005539)	0.303930 / 0.293841 (0.010090)	0.024283 / 0.128546 (-0.104263)	0.007201 / 0.075646 (-0.068446)	0.053810 / 0.419271 (-0.365461)	0.032611 / 0.043533 (-0.010922)	0.266730 / 0.255139 (0.011591)	0.281564 / 0.283200 (-0.001635)	0.018720 / 0.141683 (-0.122963)	1.140676 / 1.452155 (-0.311479)	1.206604 / 1.492716 (-0.286113)

Benchmark: benchmark_getitem_100B.json

metric	get_batch_of_1024_random_rows	get_batch_of_1024_rows	get_first_row	get_last_row
new / old (diff)	0.109390 / 0.018006 (0.091384)	0.313783 / 0.000490 (0.313294)	0.000228 / 0.000200 (0.000028)	0.000050 / 0.000054 (-0.000004)

Benchmark: benchmark_indices_mapping.json

metric	select	shard	shuffle	sort	train_test_split
new / old (diff)	0.021228 / 0.037411 (-0.016183)	0.070505 / 0.014526 (0.055979)	0.081961 / 0.176557 (-0.094595)	0.119943 / 0.737135 (-0.617193)	0.083582 / 0.296338 (-0.212757)

Benchmark: benchmark_iterating.json

metric	read 5000	read 50000	read_batch 50000 10	read_batch 50000 100	read_batch 50000 1000	read_formatted numpy 5000	read_formatted pandas 5000	read_formatted tensorflow 5000	read_formatted torch 5000	read_formatted_batch numpy 5000 10	read_formatted_batch numpy 5000 1000	shuffled read 5000	shuffled read 50000	shuffled read_batch 50000 10	shuffled read_batch 50000 100	shuffled read_batch 50000 1000	shuffled read_formatted numpy 5000	shuffled read_formatted_batch numpy 5000 10	shuffled read_formatted_batch numpy 5000 1000
new / old (diff)	0.295702 / 0.215209 (0.080493)	2.886865 / 2.077655 (0.809210)	1.583206 / 1.504120 (0.079086)	1.451129 / 1.541195 (-0.090065)	1.486253 / 1.468490 (0.017763)	0.403207 / 4.584777 (-4.181570)	2.408889 / 3.745712 (-1.336824)	2.578480 / 5.269862 (-2.691381)	1.533066 / 4.565676 (-3.032610)	0.046075 / 0.424275 (-0.378200)	0.004877 / 0.007607 (-0.002730)	0.345995 / 0.226044 (0.119950)	3.377039 / 2.268929 (1.108110)	1.944614 / 55.444624 (-53.500010)	1.677691 / 6.876477 (-5.198786)	1.672828 / 2.142072 (-0.469244)	0.468426 / 4.805227 (-4.336802)	0.097290 / 6.500664 (-6.403374)	0.040695 / 0.075469 (-0.034774)

Benchmark: benchmark_map_filter.json

metric	filter	map fast-tokenizer batched	map identity	map identity batched	map no-op batched	map no-op batched numpy	map no-op batched pandas	map no-op batched pytorch	map no-op batched tensorflow
new / old (diff)	0.965778 / 1.841788 (-0.876010)	12.092639 / 8.074308 (4.018331)	11.210968 / 10.191392 (1.019576)	0.131212 / 0.680424 (-0.549212)	0.015865 / 0.534201 (-0.518336)	0.285702 / 0.579283 (-0.293581)	0.278319 / 0.434364 (-0.156045)	0.336063 / 0.540337 (-0.204275)	0.426265 / 1.386936 (-0.960671)