Remove deprecated code

[albertvillanova]

Remove deprecated code, as part of the 3.0 release.

First merge:

• Remove metrics #6983
• Remove beam #6987
• Remove tasks #6999

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

[albertvillanova]

The removal of this code line makes apparent an underlying circular import that was introduced by:

• Better cast error when generating dataset #6509

https://github.com/huggingface/datasets/actions/runs/9694146568/job/26751140360?pr=6996

    import datasets
/opt/hostedtoolcache/Python/3.8.18/x64/lib/python3.8/site-packages/datasets/__init__.py:17: in <module>
    from .arrow_dataset import Dataset
/opt/hostedtoolcache/Python/3.8.18/x64/lib/python3.8/site-packages/datasets/arrow_dataset.py:76: in <module>
    from .arrow_reader import ArrowReader
/opt/hostedtoolcache/Python/3.8.18/x64/lib/python3.8/site-packages/datasets/arrow_reader.py:31: in <module>
    from .table import InMemoryTable, MemoryMappedTable, Table, concat_tables
/opt/hostedtoolcache/Python/3.8.18/x64/lib/python3.8/site-packages/datasets/table.py:13: in <module>
    from .utils.logging import get_logger
/opt/hostedtoolcache/Python/3.8.18/x64/lib/python3.8/site-packages/datasets/utils/__init__.py:17: in <module>
    from .info_utils import VerificationMode
/opt/hostedtoolcache/Python/3.8.18/x64/lib/python3.8/site-packages/datasets/utils/info_utils.py:8: in <module>
    from ..exceptions import (
/opt/hostedtoolcache/Python/3.8.18/x64/lib/python3.8/site-packages/datasets/exceptions.py:8: in <module>
    from .table import CastError
E   ImportError: cannot import name 'CastError' from partially initialized module 'datasets.table' (most likely due to a circular import) (/opt/hostedtoolcache/Python/3.8.18/x64/lib/python3.8/site-packages/datasets/table.py)

[albertvillanova]

I have reverted the removal of this unused import, so that the underlying circular import remain hidden.

I would propose addressing the underlying circular import in a subsequent specific 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)	0.005296 / 0.011353 (-0.006057)	0.003991 / 0.011008 (-0.007017)	0.063892 / 0.038508 (0.025384)	0.031185 / 0.023109 (0.008076)	0.248300 / 0.275898 (-0.027598)	0.270326 / 0.323480 (-0.053154)	0.004343 / 0.007986 (-0.003643)	0.002735 / 0.004328 (-0.001594)	0.049751 / 0.004250 (0.045501)	0.045629 / 0.037052 (0.008577)	0.257584 / 0.258489 (-0.000905)	0.284697 / 0.293841 (-0.009144)	0.029403 / 0.128546 (-0.099143)	0.012155 / 0.075646 (-0.063491)	0.215241 / 0.419271 (-0.204031)	0.036258 / 0.043533 (-0.007275)	0.246878 / 0.255139 (-0.008261)	0.268728 / 0.283200 (-0.014472)	0.018113 / 0.141683 (-0.123570)	1.130733 / 1.452155 (-0.321422)	1.205148 / 1.492716 (-0.287568)

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)	0.095196 / 0.018006 (0.077189)	0.300741 / 0.000490 (0.300252)	0.000220 / 0.000200 (0.000020)	0.000043 / 0.000054 (-0.000011)

Benchmark: benchmark_indices_mapping.json

metric	select	shard	shuffle	sort	train_test_split
new / old (diff)	0.018319 / 0.037411 (-0.019093)	0.062766 / 0.014526 (0.048240)	0.074748 / 0.176557 (-0.101809)	0.122177 / 0.737135 (-0.614959)	0.076652 / 0.296338 (-0.219687)

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)	0.284508 / 0.215209 (0.069299)	2.838298 / 2.077655 (0.760643)	1.480098 / 1.504120 (-0.024022)	1.362882 / 1.541195 (-0.178313)	1.389036 / 1.468490 (-0.079454)	0.747485 / 4.584777 (-3.837292)	2.385333 / 3.745712 (-1.360379)	2.924148 / 5.269862 (-2.345713)	1.869061 / 4.565676 (-2.696616)	0.079909 / 0.424275 (-0.344366)	0.005173 / 0.007607 (-0.002434)	0.345694 / 0.226044 (0.119650)	3.430648 / 2.268929 (1.161719)	1.837108 / 55.444624 (-53.607516)	1.528498 / 6.876477 (-5.347979)	1.567128 / 2.142072 (-0.574944)	0.804615 / 4.805227 (-4.000612)	0.135361 / 6.500664 (-6.365303)	0.042195 / 0.075469 (-0.033274)

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)	0.986240 / 1.841788 (-0.855548)	11.428084 / 8.074308 (3.353776)	9.168227 / 10.191392 (-1.023165)	0.131917 / 0.680424 (-0.548507)	0.014324 / 0.534201 (-0.519877)	0.302188 / 0.579283 (-0.277095)	0.263790 / 0.434364 (-0.170574)	0.343799 / 0.540337 (-0.196539)	0.428518 / 1.386936 (-0.958418)

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)	0.005734 / 0.011353 (-0.005618)	0.003914 / 0.011008 (-0.007094)	0.050105 / 0.038508 (0.011596)	0.031748 / 0.023109 (0.008639)	0.266392 / 0.275898 (-0.009506)	0.301221 / 0.323480 (-0.022259)	0.004408 / 0.007986 (-0.003578)	0.002811 / 0.004328 (-0.001517)	0.049103 / 0.004250 (0.044853)	0.041030 / 0.037052 (0.003978)	0.281003 / 0.258489 (0.022513)	0.318086 / 0.293841 (0.024245)	0.032695 / 0.128546 (-0.095852)	0.012239 / 0.075646 (-0.063408)	0.060387 / 0.419271 (-0.358885)	0.034179 / 0.043533 (-0.009354)	0.266020 / 0.255139 (0.010881)	0.288551 / 0.283200 (0.005351)	0.018778 / 0.141683 (-0.122905)	1.214959 / 1.452155 (-0.237196)	1.268269 / 1.492716 (-0.224447)

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)	0.095449 / 0.018006 (0.077443)	0.305733 / 0.000490 (0.305243)	0.000216 / 0.000200 (0.000016)	0.000052 / 0.000054 (-0.000002)

Benchmark: benchmark_indices_mapping.json

metric	select	shard	shuffle	sort	train_test_split
new / old (diff)	0.022565 / 0.037411 (-0.014847)	0.077266 / 0.014526 (0.062740)	0.089345 / 0.176557 (-0.087212)	0.128900 / 0.737135 (-0.608236)	0.089746 / 0.296338 (-0.206593)

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)	0.298221 / 0.215209 (0.083012)	2.957671 / 2.077655 (0.880016)	1.584674 / 1.504120 (0.080554)	1.456906 / 1.541195 (-0.084288)	1.467609 / 1.468490 (-0.000881)	0.718726 / 4.584777 (-3.866051)	0.948157 / 3.745712 (-2.797555)	2.953559 / 5.269862 (-2.316303)	1.895182 / 4.565676 (-2.670494)	0.078380 / 0.424275 (-0.345895)	0.005640 / 0.007607 (-0.001968)	0.352978 / 0.226044 (0.126933)	3.436341 / 2.268929 (1.167413)	1.962418 / 55.444624 (-53.482206)	1.655444 / 6.876477 (-5.221033)	1.680082 / 2.142072 (-0.461990)	0.792920 / 4.805227 (-4.012307)	0.133518 / 6.500664 (-6.367146)	0.041123 / 0.075469 (-0.034346)

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)	1.022546 / 1.841788 (-0.819242)	12.076711 / 8.074308 (4.002402)	10.159920 / 10.191392 (-0.031472)	0.143709 / 0.680424 (-0.536715)	0.015499 / 0.534201 (-0.518702)	0.302096 / 0.579283 (-0.277187)	0.125202 / 0.434364 (-0.309162)	0.349499 / 0.540337 (-0.190839)	0.456019 / 1.386936 (-0.930917)