Allow deleting a subset/config from a no-script dataset #6820

albertvillanova · 2024-04-17T14:41:12Z

TODO:

Add docs
Delete token arg from CLI example
- See: Remove token arg from CLI examples #6839

HuggingFaceDocBuilderDev · 2024-04-17T14:46:27Z

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 · 2024-04-19T12:17:18Z

This is ready for review, @huggingface/datasets.

lhoestq

Nice ! Maybe add a simple test ?

lhoestq · 2024-04-19T12:33:40Z

src/datasets/hub.py

+    operations = []
+    # data_files
+    fs = HfFileSystem(endpoint=config.HF_ENDPOINT, token=token)
+    builder = load_dataset_builder(repo_id, config_name, revision=revision, token=token)


Maybe disallow builders from a script ? you can check if the builder module starts with "datasets." for example

I have just passed trust_remote_code=False. Does it seems OK to you?

lhoestq · 2024-04-19T12:34:36Z

src/datasets/hub.py

+    fs = HfFileSystem(endpoint=config.HF_ENDPOINT, token=token)
+    builder = load_dataset_builder(repo_id, config_name, revision=revision, token=token)
+    for data_file in chain(*builder.config.data_files.values()):
+        operations.append(CommitOperationDelete(path_in_repo=fs.resolve_path(data_file).path_in_repo))


(nit) maybe add a sanity check to make sure it's a file in the same repo (in case someone defines a dataset with data_files urls from another repo)

albertvillanova · 2024-04-24T14:53:05Z

I am adding a test...

albertvillanova · 2024-04-25T12:37:00Z

@lhoestq I am getting an error in the test and I think it happens because the CI endpoint does not have the /preupload functionality:

huggingface_hub.utils._errors.RepositoryNotFoundError: 401 Client Error. (Request ID: Root=1-662a4de9-7134df595e29e4c073ac1298;332ff6e3-597a-4dfc-89df-4e9ac64215ad)

Repository Not Found for url: https://hub-ci.huggingface.co/api/datasets/__DUMMY_TRANSFORMERS_USER__/test-dataset-6c54e2-17140484441915/preupload/main?create_pr=1.
Please make sure you specified the correct `repo_id` and `repo_type`.
If you are trying to access a private or gated repo, make sure you are authenticated.
Invalid username or password.
Note: Creating a commit assumes that the repo already exists on the Huggingface Hub. Please use `create_repo` if it's not the case.

albertvillanova · 2024-04-29T12:01:54Z

@lhoestq, finally, I implemented the test with a mock of the call to HfApi.create_commit.

lhoestq

LGTM !

github-actions · 2024-04-30T09:50:37Z

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)	0.004958 / 0.011353 (-0.006395)	0.004065 / 0.011008 (-0.006943)	0.063499 / 0.038508 (0.024991)	0.030260 / 0.023109 (0.007151)	0.250910 / 0.275898 (-0.024988)	0.276632 / 0.323480 (-0.046848)	0.004038 / 0.007986 (-0.003948)	0.002721 / 0.004328 (-0.001608)	0.049098 / 0.004250 (0.044848)	0.044418 / 0.037052 (0.007366)	0.262189 / 0.258489 (0.003700)	0.292426 / 0.293841 (-0.001415)	0.027268 / 0.128546 (-0.101279)	0.010601 / 0.075646 (-0.065045)	0.207332 / 0.419271 (-0.211940)	0.036102 / 0.043533 (-0.007430)	0.252425 / 0.255139 (-0.002714)	0.269421 / 0.283200 (-0.013779)	0.018534 / 0.141683 (-0.123149)	1.127869 / 1.452155 (-0.324286)	1.179660 / 1.492716 (-0.313056)

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)	0.092686 / 0.018006 (0.074680)	0.299492 / 0.000490 (0.299002)	0.000211 / 0.000200 (0.000011)	0.000044 / 0.000054 (-0.000010)

Benchmark: benchmark_indices_mapping.json

metric	select	shard	shuffle	sort	train_test_split
new / old (diff)	0.018385 / 0.037411 (-0.019026)	0.060979 / 0.014526 (0.046453)	0.073351 / 0.176557 (-0.103205)	0.120145 / 0.737135 (-0.616990)	0.073653 / 0.296338 (-0.222686)

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)	0.286175 / 0.215209 (0.070966)	2.792698 / 2.077655 (0.715043)	1.507442 / 1.504120 (0.003322)	1.392531 / 1.541195 (-0.148664)	1.387253 / 1.468490 (-0.081237)	0.568435 / 4.584777 (-4.016342)	2.387392 / 3.745712 (-1.358321)	2.813695 / 5.269862 (-2.456167)	1.747392 / 4.565676 (-2.818284)	0.062948 / 0.424275 (-0.361328)	0.005596 / 0.007607 (-0.002011)	0.334357 / 0.226044 (0.108313)	3.263289 / 2.268929 (0.994360)	1.829553 / 55.444624 (-53.615071)	1.552510 / 6.876477 (-5.323967)	1.579975 / 2.142072 (-0.562098)	0.633982 / 4.805227 (-4.171246)	0.118752 / 6.500664 (-6.381912)	0.042445 / 0.075469 (-0.033024)

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)	0.988062 / 1.841788 (-0.853725)	11.615693 / 8.074308 (3.541385)	9.728103 / 10.191392 (-0.463289)	0.131561 / 0.680424 (-0.548862)	0.015330 / 0.534201 (-0.518871)	0.289617 / 0.579283 (-0.289666)	0.265717 / 0.434364 (-0.168646)	0.323974 / 0.540337 (-0.216363)	0.419523 / 1.386936 (-0.967413)

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)	0.005385 / 0.011353 (-0.005968)	0.003753 / 0.011008 (-0.007255)	0.049821 / 0.038508 (0.011313)	0.030490 / 0.023109 (0.007381)	0.260550 / 0.275898 (-0.015348)	0.284598 / 0.323480 (-0.038881)	0.004165 / 0.007986 (-0.003821)	0.002741 / 0.004328 (-0.001588)	0.048567 / 0.004250 (0.044317)	0.045185 / 0.037052 (0.008133)	0.273164 / 0.258489 (0.014674)	0.301995 / 0.293841 (0.008155)	0.028802 / 0.128546 (-0.099744)	0.010539 / 0.075646 (-0.065108)	0.057967 / 0.419271 (-0.361305)	0.032826 / 0.043533 (-0.010706)	0.260425 / 0.255139 (0.005286)	0.280175 / 0.283200 (-0.003024)	0.017202 / 0.141683 (-0.124481)	1.129588 / 1.452155 (-0.322567)	1.199565 / 1.492716 (-0.293152)

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)	0.091234 / 0.018006 (0.073228)	0.299313 / 0.000490 (0.298824)	0.000203 / 0.000200 (0.000003)	0.000044 / 0.000054 (-0.000010)

Benchmark: benchmark_indices_mapping.json

metric	select	shard	shuffle	sort	train_test_split
new / old (diff)	0.022519 / 0.037411 (-0.014892)	0.075915 / 0.014526 (0.061389)	0.088636 / 0.176557 (-0.087920)	0.128234 / 0.737135 (-0.608902)	0.089782 / 0.296338 (-0.206556)

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)	0.291936 / 0.215209 (0.076727)	2.864589 / 2.077655 (0.786935)	1.575649 / 1.504120 (0.071529)	1.452797 / 1.541195 (-0.088398)	1.476245 / 1.468490 (0.007754)	0.593972 / 4.584777 (-3.990804)	0.962315 / 3.745712 (-2.783397)	2.836496 / 5.269862 (-2.433366)	1.758639 / 4.565676 (-2.807038)	0.064842 / 0.424275 (-0.359433)	0.005076 / 0.007607 (-0.002531)	0.342568 / 0.226044 (0.116524)	3.392753 / 2.268929 (1.123825)	1.908305 / 55.444624 (-53.536319)	1.632140 / 6.876477 (-5.244337)	1.653048 / 2.142072 (-0.489024)	0.662068 / 4.805227 (-4.143159)	0.118326 / 6.500664 (-6.382338)	0.041222 / 0.075469 (-0.034247)

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)	1.005119 / 1.841788 (-0.836669)	12.250922 / 8.074308 (4.176614)	9.775600 / 10.191392 (-0.415792)	0.146230 / 0.680424 (-0.534194)	0.015883 / 0.534201 (-0.518318)	0.290807 / 0.579283 (-0.288476)	0.126002 / 0.434364 (-0.308362)	0.392332 / 0.540337 (-0.148005)	0.435513 / 1.386936 (-0.951423)

Add delete_from_hub function

16baa91

albertvillanova added 4 commits April 18, 2024 11:58

Fix for None config_names

711adcc

Fix for None configs or dataset_info

a466530

Create always a pull request

fea18b3

Add CLI delete_from_hub command

287ff3a

albertvillanova marked this pull request as ready for review April 19, 2024 11:01

lhoestq reviewed Apr 19, 2024

View reviewed changes

albertvillanova added 2 commits April 24, 2024 16:38

Pass trust_remote_code False

d4ac62f

Delete only data files in the same repo

efb5bd4

albertvillanova added 8 commits April 25, 2024 14:43

Merge remote-tracking branch 'upstream/main' into fix-6810

ecec1a3

Make config_name arg required

cb8b130

Update help message for CLI dataset_id arg

ab8778b

Update docs about the CLI

255754e

Add docstring

73c46e1

Merge remote-tracking branch 'upstream/main' into fix-6810

89bc37c

Add example and tip to log in

3c48262

Test delete_fro_hub

efb6f7a

albertvillanova requested a review from lhoestq April 30, 2024 08:52

lhoestq approved these changes Apr 30, 2024

View reviewed changes

albertvillanova merged commit ceb25e1 into main Apr 30, 2024
12 checks passed

albertvillanova deleted the fix-6810 branch April 30, 2024 09:44

albertvillanova mentioned this pull request May 2, 2024

CI fails on Windows for test_delete_from_hub and test_xgetsize_private due to new-line character #6856

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Allow deleting a subset/config from a no-script dataset #6820

Allow deleting a subset/config from a no-script dataset #6820

albertvillanova commented Apr 17, 2024 •

edited

Loading

HuggingFaceDocBuilderDev commented Apr 17, 2024

albertvillanova commented Apr 19, 2024

lhoestq left a comment

lhoestq Apr 19, 2024

albertvillanova Apr 24, 2024

lhoestq Apr 19, 2024

albertvillanova Apr 24, 2024

albertvillanova commented Apr 24, 2024

albertvillanova commented Apr 25, 2024

albertvillanova commented Apr 29, 2024

lhoestq left a comment

github-actions bot commented Apr 30, 2024

Benchmark: benchmark_array_xd.json

Benchmark: benchmark_getitem_100B.json

Benchmark: benchmark_indices_mapping.json

Benchmark: benchmark_iterating.json

Benchmark: benchmark_map_filter.json

Benchmark: benchmark_array_xd.json

Benchmark: benchmark_getitem_100B.json

Benchmark: benchmark_indices_mapping.json

Benchmark: benchmark_iterating.json

Benchmark: benchmark_map_filter.json

Allow deleting a subset/config from a no-script dataset #6820

Allow deleting a subset/config from a no-script dataset #6820

Conversation

albertvillanova commented Apr 17, 2024 • edited Loading

HuggingFaceDocBuilderDev commented Apr 17, 2024

albertvillanova commented Apr 19, 2024

lhoestq left a comment

Choose a reason for hiding this comment

lhoestq Apr 19, 2024

Choose a reason for hiding this comment

albertvillanova Apr 24, 2024

Choose a reason for hiding this comment

lhoestq Apr 19, 2024

Choose a reason for hiding this comment

albertvillanova Apr 24, 2024

Choose a reason for hiding this comment

albertvillanova commented Apr 24, 2024

albertvillanova commented Apr 25, 2024

albertvillanova commented Apr 29, 2024

lhoestq left a comment

Choose a reason for hiding this comment

github-actions bot commented Apr 30, 2024

Benchmark: benchmark_array_xd.json

Benchmark: benchmark_getitem_100B.json

Benchmark: benchmark_indices_mapping.json

Benchmark: benchmark_iterating.json

Benchmark: benchmark_map_filter.json

Benchmark: benchmark_array_xd.json

Benchmark: benchmark_getitem_100B.json

Benchmark: benchmark_indices_mapping.json

Benchmark: benchmark_iterating.json

Benchmark: benchmark_map_filter.json

albertvillanova commented Apr 17, 2024 •

edited

Loading