Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Unify load_from_cache_file type and logic #5515

Merged
merged 2 commits into from
Feb 14, 2023
Merged

Unify load_from_cache_file type and logic #5515

merged 2 commits into from
Feb 14, 2023

Conversation

HallerPatrick
Copy link
Contributor

  • Updating type annotations for #load_from_cache_file
  • Added logic for cache checking if needed
  • Updated documentation following the wording of Dataset.map

@HallerPatrick
* Updating type annotations for load_from_cache_file
416369b 
* Added logic for cache checking if needed
* Updated documentation following the wording of `Dataset.map`
@HallerPatrick HallerPatrick mentioned this pull request Feb 9, 2023
Closed
@HuggingFaceDocBuilderDev
Copy link

HuggingFaceDocBuilderDev commented Feb 9, 2023 •
edited
Loading

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

@mariosasko mariosasko linked an issue Feb 9, 2023 that may be closed by this pull request
Improve inconsistency of Dataset.map interface for load_from_cache_file #5514
Closed
mariosasko
mariosasko reviewed Feb 9, 2023
View reviewed changes
Copy link
Collaborator

@mariosasko mariosasko left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Good job, thanks! DatasetDict's methods also need to be updated with the new default value, and then we can merge.

@HallerPatrick
Copy link
Contributor Author

The commit also includes the changes to the DatasetDict methods or am I missing something?

@mariosasko
Copy link
Collaborator

Oh, indeed. Feel free to mark the PR as "Ready for review" then.

@HallerPatrick HallerPatrick marked this pull request as ready for review February 9, 2023 15:48
mariosasko
mariosasko approved these changes Feb 9, 2023
View reviewed changes
Copy link
Collaborator

@mariosasko mariosasko left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Nice, lgtm!

lhoestq
lhoestq reviewed Feb 12, 2023
View reviewed changes
Copy link
Member

@lhoestq lhoestq left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks ! Just one comment:

src/datasets/arrow_dataset.py Outdated Show resolved Hide resolved
@mariosasko mariosasko merged commit 819bc6e into huggingface:main Feb 14, 2023
@github-actions
Copy link

Show benchmarks

PyArrow==6.0.0

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) 0.010149 / 0.011353 (-0.001204) 0.005606 / 0.011008 (-0.005402) 0.103455 / 0.038508 (0.064947) 0.042934 / 0.023109 (0.019825) 0.308365 / 0.275898 (0.032467) 0.394188 / 0.323480 (0.070708) 0.008760 / 0.007986 (0.000774) 0.004567 / 0.004328 (0.000239) 0.077959 / 0.004250 (0.073708) 0.050115 / 0.037052 (0.013063) 0.318009 / 0.258489 (0.059520) 0.358578 / 0.293841 (0.064737) 0.039231 / 0.128546 (-0.089315) 0.012381 / 0.075646 (-0.063265) 0.340046 / 0.419271 (-0.079226) 0.048366 / 0.043533 (0.004834) 0.307643 / 0.255139 (0.052504) 0.342886 / 0.283200 (0.059687) 0.109628 / 0.141683 (-0.032055) 1.457297 / 1.452155 (0.005142) 1.518067 / 1.492716 (0.025351)

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) 0.295590 / 0.018006 (0.277584) 0.531515 / 0.000490 (0.531026) 0.005677 / 0.000200 (0.005477) 0.000095 / 0.000054 (0.000041)

Benchmark: benchmark_indices_mapping.json

metric select shard shuffle sort train_test_split
new / old (diff) 0.030901 / 0.037411 (-0.006511) 0.118312 / 0.014526 (0.103786) 0.123146 / 0.176557 (-0.053410) 0.163608 / 0.737135 (-0.573527) 0.128604 / 0.296338 (-0.167734)

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) 0.404143 / 0.215209 (0.188934) 4.000118 / 2.077655 (1.922464) 1.804502 / 1.504120 (0.300382) 1.597287 / 1.541195 (0.056093) 1.738512 / 1.468490 (0.270022) 0.704658 / 4.584777 (-3.880119) 3.830101 / 3.745712 (0.084389) 2.186598 / 5.269862 (-3.083263) 1.367873 / 4.565676 (-3.197804) 0.085550 / 0.424275 (-0.338725) 0.012226 / 0.007607 (0.004619) 0.505760 / 0.226044 (0.279716) 5.054583 / 2.268929 (2.785655) 2.284942 / 55.444624 (-53.159682) 1.961413 / 6.876477 (-4.915064) 2.059449 / 2.142072 (-0.082623) 0.845009 / 4.805227 (-3.960218) 0.167204 / 6.500664 (-6.333460) 0.065998 / 0.075469 (-0.009471)

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) 1.221861 / 1.841788 (-0.619927) 15.925213 / 8.074308 (7.850905) 15.359308 / 10.191392 (5.167916) 0.171776 / 0.680424 (-0.508648) 0.029234 / 0.534201 (-0.504967) 0.446349 / 0.579283 (-0.132934) 0.447873 / 0.434364 (0.013509) 0.527400 / 0.540337 (-0.012937) 0.610208 / 1.386936 (-0.776728)
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) 0.008030 / 0.011353 (-0.003323) 0.005686 / 0.011008 (-0.005322) 0.076204 / 0.038508 (0.037696) 0.037131 / 0.023109 (0.014022) 0.341461 / 0.275898 (0.065563) 0.378734 / 0.323480 (0.055255) 0.006580 / 0.007986 (-0.001406) 0.004379 / 0.004328 (0.000050) 0.073983 / 0.004250 (0.069732) 0.055895 / 0.037052 (0.018842) 0.342667 / 0.258489 (0.084178) 0.401464 / 0.293841 (0.107623) 0.037710 / 0.128546 (-0.090837) 0.012604 / 0.075646 (-0.063042) 0.087563 / 0.419271 (-0.331709) 0.050887 / 0.043533 (0.007354) 0.333491 / 0.255139 (0.078352) 0.357437 / 0.283200 (0.074237) 0.109566 / 0.141683 (-0.032117) 1.423372 / 1.452155 (-0.028783) 1.569423 / 1.492716 (0.076706)

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) 0.340986 / 0.018006 (0.322980) 0.530885 / 0.000490 (0.530395) 0.004172 / 0.000200 (0.003972) 0.000115 / 0.000054 (0.000060)

Benchmark: benchmark_indices_mapping.json

metric select shard shuffle sort train_test_split
new / old (diff) 0.030424 / 0.037411 (-0.006987) 0.121191 / 0.014526 (0.106666) 0.129066 / 0.176557 (-0.047491) 0.166938 / 0.737135 (-0.570198) 0.132000 / 0.296338 (-0.164338)

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) 0.418718 / 0.215209 (0.203509) 4.163973 / 2.077655 (2.086318) 1.982665 / 1.504120 (0.478545) 1.798866 / 1.541195 (0.257671) 1.918867 / 1.468490 (0.450377) 0.724634 / 4.584777 (-3.860143) 3.864549 / 3.745712 (0.118837) 3.697768 / 5.269862 (-1.572093) 1.983942 / 4.565676 (-2.581735) 0.086818 / 0.424275 (-0.337457) 0.012336 / 0.007607 (0.004728) 0.522314 / 0.226044 (0.296269) 5.216813 / 2.268929 (2.947884) 2.516187 / 55.444624 (-52.928437) 2.172057 / 6.876477 (-4.704420) 2.342773 / 2.142072 (0.200701) 0.851805 / 4.805227 (-3.953422) 0.170139 / 6.500664 (-6.330525) 0.068494 / 0.075469 (-0.006975)

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) 1.307370 / 1.841788 (-0.534418) 16.737937 / 8.074308 (8.663629) 14.483384 / 10.191392 (4.291992) 0.172418 / 0.680424 (-0.508006) 0.018241 / 0.534201 (-0.515960) 0.432049 / 0.579283 (-0.147234) 0.447590 / 0.434364 (0.013227) 0.550332 / 0.540337 (0.009994) 0.646756 / 1.386936 (-0.740180)

@HallerPatrick HallerPatrick deleted the fix-load_from_cache_file_default_value branch February 14, 2023 15:38
AJDERS pushed a commit to AJDERS/datasets that referenced this pull request Feb 15, 2023
@HallerPatrick @mariosasko @AJDERS
Unify load_from_cache_file type and logic (huggingface#5515)
1223779 
* Updating type annotations for `load_from_cache_file`
* Added logic for cache checking if needed
* Updated documentation following the wording of `Dataset.map`

Co-authored-by: Mario Šaško <mariosasko777@gmail.com>
filip-halt pushed a commit to filip-halt/datasets that referenced this pull request Feb 16, 2023
@HallerPatrick @mariosasko
Unify load_from_cache_file type and logic (huggingface#5515)
f862099 
* Updating type annotations for `load_from_cache_file`
* Added logic for cache checking if needed
* Updated documentation following the wording of `Dataset.map`

Co-authored-by: Mario Šaško <mariosasko777@gmail.com>
filip-halt pushed a commit to filip-halt/datasets that referenced this pull request Feb 16, 2023
@HallerPatrick @mariosasko
Unify load_from_cache_file type and logic (huggingface#5515)
65bad7e 
* Updating type annotations for `load_from_cache_file`
* Added logic for cache checking if needed
* Updated documentation following the wording of `Dataset.map`

Co-authored-by: Mario Šaško <mariosasko777@gmail.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@lhoestq lhoestq lhoestq left review comments

@mariosasko mariosasko mariosasko approved these changes

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.

Improve inconsistency of Dataset.map interface for load_from_cache_file
4 participants
@HallerPatrick @HuggingFaceDocBuilderDev @mariosasko @lhoestq