Make convert_to_parquet CLI command create script branch

[albertvillanova]

Make convert_to_parquet CLI command create a "script" branch and keep the script file on it.

This PR proposes the simplest UX approach: whenever --revision is not explicitly passed (i.e., when the script is in the main branch), try to create a "script" branch from the "main" branch; if the "script" branch exists already, then do nothing.

Follow-up of:

• Add CLI function to convert script-dataset to Parquet #6795

Close #6808.

CC: @severo

[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]

@huggingface/datasets once this PR is merged, I would suggest making a release. Do you agree?

• This PR is a follow-up of Add CLI function to convert script-dataset to Parquet #6795

[lhoestq]

LGTM ! and yes let's do a release once it's merged

(also making sure the transformers CI is green with datasets@main before the release - we had many small changes this time)

[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)	0.004963 / 0.011353 (-0.006390)	0.003121 / 0.011008 (-0.007888)	0.063421 / 0.038508 (0.024913)	0.030727 / 0.023109 (0.007618)	0.237698 / 0.275898 (-0.038200)	0.266613 / 0.323480 (-0.056867)	0.004237 / 0.007986 (-0.003749)	0.002715 / 0.004328 (-0.001614)	0.049503 / 0.004250 (0.045253)	0.043705 / 0.037052 (0.006653)	0.247818 / 0.258489 (-0.010671)	0.287545 / 0.293841 (-0.006296)	0.027232 / 0.128546 (-0.101314)	0.009952 / 0.075646 (-0.065695)	0.208678 / 0.419271 (-0.210593)	0.035494 / 0.043533 (-0.008039)	0.260900 / 0.255139 (0.005761)	0.264738 / 0.283200 (-0.018461)	0.018093 / 0.141683 (-0.123590)	1.130924 / 1.452155 (-0.321231)	1.178982 / 1.492716 (-0.313734)

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)	0.094610 / 0.018006 (0.076604)	0.304674 / 0.000490 (0.304184)	0.000215 / 0.000200 (0.000015)	0.000048 / 0.000054 (-0.000007)

Benchmark: benchmark_indices_mapping.json

metric	select	shard	shuffle	sort	train_test_split
new / old (diff)	0.018168 / 0.037411 (-0.019243)	0.062040 / 0.014526 (0.047514)	0.075634 / 0.176557 (-0.100922)	0.119488 / 0.737135 (-0.617647)	0.074790 / 0.296338 (-0.221548)

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)	0.282449 / 0.215209 (0.067240)	2.773231 / 2.077655 (0.695576)	1.455156 / 1.504120 (-0.048964)	1.332652 / 1.541195 (-0.208543)	1.340795 / 1.468490 (-0.127695)	0.576588 / 4.584777 (-4.008189)	2.415513 / 3.745712 (-1.330199)	2.801569 / 5.269862 (-2.468292)	1.741039 / 4.565676 (-2.824637)	0.064386 / 0.424275 (-0.359890)	0.005293 / 0.007607 (-0.002314)	0.329732 / 0.226044 (0.103688)	3.227275 / 2.268929 (0.958347)	1.793121 / 55.444624 (-53.651503)	1.515115 / 6.876477 (-5.361362)	1.518738 / 2.142072 (-0.623335)	0.664465 / 4.805227 (-4.140762)	0.118813 / 6.500664 (-6.381851)	0.041715 / 0.075469 (-0.033754)

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)	0.974371 / 1.841788 (-0.867416)	11.432869 / 8.074308 (3.358561)	9.607939 / 10.191392 (-0.583453)	0.143996 / 0.680424 (-0.536427)	0.014624 / 0.534201 (-0.519577)	0.286899 / 0.579283 (-0.292384)	0.265965 / 0.434364 (-0.168399)	0.324727 / 0.540337 (-0.215611)	0.420917 / 1.386936 (-0.966019)

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)	0.005145 / 0.011353 (-0.006207)	0.003723 / 0.011008 (-0.007286)	0.050387 / 0.038508 (0.011879)	0.030734 / 0.023109 (0.007625)	0.274331 / 0.275898 (-0.001567)	0.295045 / 0.323480 (-0.028435)	0.004187 / 0.007986 (-0.003799)	0.002781 / 0.004328 (-0.001547)	0.049698 / 0.004250 (0.045448)	0.040049 / 0.037052 (0.002996)	0.284016 / 0.258489 (0.025527)	0.309908 / 0.293841 (0.016067)	0.028994 / 0.128546 (-0.099552)	0.010625 / 0.075646 (-0.065021)	0.059305 / 0.419271 (-0.359967)	0.032982 / 0.043533 (-0.010551)	0.273342 / 0.255139 (0.018203)	0.291726 / 0.283200 (0.008527)	0.018084 / 0.141683 (-0.123599)	1.136864 / 1.452155 (-0.315290)	1.163656 / 1.492716 (-0.329061)

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)	0.094868 / 0.018006 (0.076862)	0.302900 / 0.000490 (0.302410)	0.000226 / 0.000200 (0.000026)	0.000053 / 0.000054 (-0.000002)

Benchmark: benchmark_indices_mapping.json

metric	select	shard	shuffle	sort	train_test_split
new / old (diff)	0.022142 / 0.037411 (-0.015269)	0.077457 / 0.014526 (0.062932)	0.087989 / 0.176557 (-0.088568)	0.127354 / 0.737135 (-0.609781)	0.092027 / 0.296338 (-0.204312)

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)	0.291196 / 0.215209 (0.075987)	2.840386 / 2.077655 (0.762731)	1.571201 / 1.504120 (0.067081)	1.449429 / 1.541195 (-0.091765)	1.467189 / 1.468490 (-0.001301)	0.580991 / 4.584777 (-4.003786)	2.422566 / 3.745712 (-1.323146)	2.839621 / 5.269862 (-2.430240)	1.782987 / 4.565676 (-2.782689)	0.064765 / 0.424275 (-0.359510)	0.005338 / 0.007607 (-0.002269)	0.349148 / 0.226044 (0.123104)	3.421283 / 2.268929 (1.152355)	1.943503 / 55.444624 (-53.501122)	1.653881 / 6.876477 (-5.222596)	1.698141 / 2.142072 (-0.443931)	0.667628 / 4.805227 (-4.137599)	0.118469 / 6.500664 (-6.382195)	0.041693 / 0.075469 (-0.033776)

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)	1.026385 / 1.841788 (-0.815403)	12.225049 / 8.074308 (4.150741)	10.363072 / 10.191392 (0.171680)	0.142682 / 0.680424 (-0.537742)	0.015698 / 0.534201 (-0.518502)	0.288148 / 0.579283 (-0.291135)	0.272639 / 0.434364 (-0.161724)	0.325305 / 0.540337 (-0.215032)	0.421395 / 1.386936 (-0.965541)