Rename LargeList.dtype to LargeList.feature

[albertvillanova]

Rename LargeList.dtype to LargeList.feature.

Note that dtype is usually used for NumPy data types ("int64", "float32",...): see Value.dtype.

However, LargeList attribute (like Sequence.feature) expects a FeatureType instead.

With this renaming:

• we avoid confusion about the expected type and
• we also align LargeList with Sequence.

[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.

[lhoestq]

good idea

[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)	0.005598 / 0.011353 (-0.005755)	0.004327 / 0.011008 (-0.006681)	0.063961 / 0.038508 (0.025453)	0.031039 / 0.023109 (0.007930)	0.245586 / 0.275898 (-0.030312)	0.273765 / 0.323480 (-0.049715)	0.003463 / 0.007986 (-0.004523)	0.002871 / 0.004328 (-0.001457)	0.049169 / 0.004250 (0.044918)	0.049342 / 0.037052 (0.012290)	0.259255 / 0.258489 (0.000766)	0.295688 / 0.293841 (0.001847)	0.029527 / 0.128546 (-0.099019)	0.012507 / 0.075646 (-0.063139)	0.209420 / 0.419271 (-0.209851)	0.036666 / 0.043533 (-0.006866)	0.272031 / 0.255139 (0.016892)	0.272585 / 0.283200 (-0.010614)	0.020004 / 0.141683 (-0.121679)	1.158605 / 1.452155 (-0.293550)	1.230930 / 1.492716 (-0.261787)

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)	0.109196 / 0.018006 (0.091189)	0.377759 / 0.000490 (0.377270)	0.000222 / 0.000200 (0.000022)	0.000043 / 0.000054 (-0.000011)

Benchmark: benchmark_indices_mapping.json

metric	select	shard	shuffle	sort	train_test_split
new / old (diff)	0.018961 / 0.037411 (-0.018450)	0.063189 / 0.014526 (0.048663)	0.075253 / 0.176557 (-0.101303)	0.122912 / 0.737135 (-0.614223)	0.077961 / 0.296338 (-0.218378)

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)	0.278425 / 0.215209 (0.063216)	2.748336 / 2.077655 (0.670681)	1.468410 / 1.504120 (-0.035710)	1.347859 / 1.541195 (-0.193336)	1.389175 / 1.468490 (-0.079315)	0.742833 / 4.584777 (-3.841943)	2.358930 / 3.745712 (-1.386782)	3.062720 / 5.269862 (-2.207141)	1.912264 / 4.565676 (-2.653412)	0.079263 / 0.424275 (-0.345012)	0.005212 / 0.007607 (-0.002396)	0.332482 / 0.226044 (0.106438)	3.287045 / 2.268929 (1.018116)	1.827862 / 55.444624 (-53.616762)	1.525087 / 6.876477 (-5.351390)	1.581742 / 2.142072 (-0.560330)	0.791737 / 4.805227 (-4.013490)	0.135774 / 6.500664 (-6.364890)	0.043700 / 0.075469 (-0.031769)

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)	0.982104 / 1.841788 (-0.859683)	12.227639 / 8.074308 (4.153331)	9.492719 / 10.191392 (-0.698673)	0.144792 / 0.680424 (-0.535632)	0.014844 / 0.534201 (-0.519357)	0.304919 / 0.579283 (-0.274364)	0.262955 / 0.434364 (-0.171409)	0.339517 / 0.540337 (-0.200821)	0.430929 / 1.386936 (-0.956007)

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)	0.005982 / 0.011353 (-0.005371)	0.004199 / 0.011008 (-0.006809)	0.050674 / 0.038508 (0.012166)	0.032713 / 0.023109 (0.009604)	0.270071 / 0.275898 (-0.005827)	0.300469 / 0.323480 (-0.023011)	0.005159 / 0.007986 (-0.002826)	0.002961 / 0.004328 (-0.001368)	0.048403 / 0.004250 (0.044152)	0.042024 / 0.037052 (0.004971)	0.288927 / 0.258489 (0.030438)	0.321412 / 0.293841 (0.027571)	0.032436 / 0.128546 (-0.096110)	0.012472 / 0.075646 (-0.063175)	0.060527 / 0.419271 (-0.358744)	0.034222 / 0.043533 (-0.009311)	0.276259 / 0.255139 (0.021120)	0.293168 / 0.283200 (0.009969)	0.019245 / 0.141683 (-0.122438)	1.180766 / 1.452155 (-0.271388)	1.220269 / 1.492716 (-0.272447)

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)	0.110082 / 0.018006 (0.092076)	0.364221 / 0.000490 (0.363731)	0.000221 / 0.000200 (0.000021)	0.000044 / 0.000054 (-0.000011)

Benchmark: benchmark_indices_mapping.json

metric	select	shard	shuffle	sort	train_test_split
new / old (diff)	0.022923 / 0.037411 (-0.014488)	0.078022 / 0.014526 (0.063496)	0.089543 / 0.176557 (-0.087013)	0.129855 / 0.737135 (-0.607280)	0.090891 / 0.296338 (-0.205448)

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)	0.304169 / 0.215209 (0.088960)	2.969772 / 2.077655 (0.892117)	1.582647 / 1.504120 (0.078527)	1.464446 / 1.541195 (-0.076749)	1.485422 / 1.468490 (0.016932)	0.720105 / 4.584777 (-3.864672)	0.966730 / 3.745712 (-2.778982)	3.017549 / 5.269862 (-2.252313)	1.924574 / 4.565676 (-2.641103)	0.079938 / 0.424275 (-0.344337)	0.005684 / 0.007607 (-0.001923)	0.364093 / 0.226044 (0.138048)	3.569470 / 2.268929 (1.300541)	1.956535 / 55.444624 (-53.488089)	1.669432 / 6.876477 (-5.207045)	1.687596 / 2.142072 (-0.454476)	0.802725 / 4.805227 (-4.002502)	0.132874 / 6.500664 (-6.367790)	0.041403 / 0.075469 (-0.034067)

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)	1.033317 / 1.841788 (-0.808471)	12.590652 / 8.074308 (4.516344)	10.618609 / 10.191392 (0.427217)	0.131833 / 0.680424 (-0.548591)	0.015675 / 0.534201 (-0.518526)	0.300804 / 0.579283 (-0.278479)	0.127253 / 0.434364 (-0.307111)	0.342559 / 0.540337 (-0.197779)	0.464302 / 1.386936 (-0.922634)