Skip to content

Commit

Permalink
Release: 1.18.3
Browse files Browse the repository at this point in the history
  • Loading branch information
lhoestq committed Feb 2, 2022
1 parent 99460c5 commit c6bc52a
Show file tree
Hide file tree
Showing 3 changed files with 3 additions and 3 deletions.
2 changes: 1 addition & 1 deletion docs/source/conf.py
Original file line number Diff line number Diff line change
Expand Up @@ -25,7 +25,7 @@
# The short X.Y version
version = ""
# The full version, including alpha/beta/rc tags
release = "1.18.2"
release = "1.18.3"


# -- General configuration ---------------------------------------------------
Expand Down
2 changes: 1 addition & 1 deletion setup.py
Original file line number Diff line number Diff line change
Expand Up @@ -228,7 +228,7 @@

setup(
name="datasets",
version="1.18.3.dev0", # expected format is one of x.y.z.dev0, or x.y.z.rc1 or x.y.z (no to dashes, yes to dots)
version="1.18.3", # expected format is one of x.y.z.dev0, or x.y.z.rc1 or x.y.z (no to dashes, yes to dots)
description="HuggingFace community-driven open-source library of datasets",
long_description=open("README.md", encoding="utf-8").read(),
long_description_content_type="text/markdown",
Expand Down
2 changes: 1 addition & 1 deletion src/datasets/__init__.py
Original file line number Diff line number Diff line change
Expand Up @@ -17,7 +17,7 @@
# pylint: enable=line-too-long
# pylint: disable=g-import-not-at-top,g-bad-import-order,wrong-import-position

__version__ = "1.18.3.dev0"
__version__ = "1.18.3"

import pyarrow
from packaging import version as _version
Expand Down

3 comments on commit c6bc52a

@github-actions
Copy link

Choose a reason for hiding this comment

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

Show benchmarks

PyArrow==3.0.0

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) 0.010044 / 0.011353 (-0.001309) 0.004070 / 0.011008 (-0.006938) 0.031243 / 0.038508 (-0.007265) 0.035379 / 0.023109 (0.012270) 0.295964 / 0.275898 (0.020066) 0.334450 / 0.323480 (0.010970) 0.008038 / 0.007986 (0.000052) 0.003757 / 0.004328 (-0.000571) 0.009167 / 0.004250 (0.004916) 0.045200 / 0.037052 (0.008147) 0.288684 / 0.258489 (0.030195) 0.334510 / 0.293841 (0.040669) 0.031734 / 0.128546 (-0.096812) 0.009830 / 0.075646 (-0.065816) 0.253621 / 0.419271 (-0.165651) 0.051500 / 0.043533 (0.007968) 0.292722 / 0.255139 (0.037583) 0.316949 / 0.283200 (0.033749) 0.109218 / 0.141683 (-0.032465) 1.872118 / 1.452155 (0.419963) 1.913131 / 1.492716 (0.420415)

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) 0.324184 / 0.018006 (0.306178) 0.511944 / 0.000490 (0.511454) 0.024825 / 0.000200 (0.024625) 0.000389 / 0.000054 (0.000335)

Benchmark: benchmark_indices_mapping.json

metric select shard shuffle sort train_test_split
new / old (diff) 0.037581 / 0.037411 (0.000170) 0.022596 / 0.014526 (0.008070) 0.034612 / 0.176557 (-0.141945) 0.075454 / 0.737135 (-0.661681) 0.034774 / 0.296338 (-0.261564)

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) 0.413462 / 0.215209 (0.198253) 4.130333 / 2.077655 (2.052678) 1.770515 / 1.504120 (0.266395) 1.567501 / 1.541195 (0.026306) 1.681172 / 1.468490 (0.212682) 0.438749 / 4.584777 (-4.146028) 4.604207 / 3.745712 (0.858495) 3.948200 / 5.269862 (-1.321662) 0.924135 / 4.565676 (-3.641541) 0.053021 / 0.424275 (-0.371254) 0.012225 / 0.007607 (0.004618) 0.519443 / 0.226044 (0.293399) 5.176187 / 2.268929 (2.907259) 2.268276 / 55.444624 (-53.176348) 1.890657 / 6.876477 (-4.985820) 1.949728 / 2.142072 (-0.192345) 0.555071 / 4.805227 (-4.250157) 0.123256 / 6.500664 (-6.377408) 0.062849 / 0.075469 (-0.012621)

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) 1.594929 / 1.841788 (-0.246858) 14.715473 / 8.074308 (6.641165) 26.696118 / 10.191392 (16.504726) 0.880268 / 0.680424 (0.199844) 0.514483 / 0.534201 (-0.019718) 0.503179 / 0.579283 (-0.076104) 0.506234 / 0.434364 (0.071870) 0.324792 / 0.540337 (-0.215545) 0.337031 / 1.386936 (-1.049905)
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) 0.008304 / 0.011353 (-0.003049) 0.003944 / 0.011008 (-0.007065) 0.029477 / 0.038508 (-0.009031) 0.034199 / 0.023109 (0.011090) 0.303434 / 0.275898 (0.027536) 0.327454 / 0.323480 (0.003974) 0.006309 / 0.007986 (-0.001677) 0.004925 / 0.004328 (0.000596) 0.007355 / 0.004250 (0.003104) 0.040232 / 0.037052 (0.003180) 0.293051 / 0.258489 (0.034562) 0.325613 / 0.293841 (0.031772) 0.031574 / 0.128546 (-0.096972) 0.009732 / 0.075646 (-0.065915) 0.252014 / 0.419271 (-0.167258) 0.050424 / 0.043533 (0.006892) 0.294450 / 0.255139 (0.039312) 0.319943 / 0.283200 (0.036744) 0.092940 / 0.141683 (-0.048742) 1.750508 / 1.452155 (0.298354) 1.801591 / 1.492716 (0.308874)

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) 0.373015 / 0.018006 (0.355009) 0.513520 / 0.000490 (0.513030) 0.040155 / 0.000200 (0.039955) 0.000683 / 0.000054 (0.000628)

Benchmark: benchmark_indices_mapping.json

metric select shard shuffle sort train_test_split
new / old (diff) 0.033647 / 0.037411 (-0.003765) 0.021590 / 0.014526 (0.007064) 0.032045 / 0.176557 (-0.144511) 0.082389 / 0.737135 (-0.654746) 0.030275 / 0.296338 (-0.266063)

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) 0.419589 / 0.215209 (0.204380) 4.208030 / 2.077655 (2.130375) 1.803310 / 1.504120 (0.299190) 1.594332 / 1.541195 (0.053137) 1.674219 / 1.468490 (0.205728) 0.438966 / 4.584777 (-4.145811) 4.548299 / 3.745712 (0.802587) 2.160592 / 5.269862 (-3.109269) 0.911965 / 4.565676 (-3.653711) 0.053180 / 0.424275 (-0.371096) 0.012570 / 0.007607 (0.004962) 0.528195 / 0.226044 (0.302151) 5.265138 / 2.268929 (2.996210) 2.273141 / 55.444624 (-53.171484) 1.877421 / 6.876477 (-4.999056) 2.046718 / 2.142072 (-0.095355) 0.565038 / 4.805227 (-4.240190) 0.122719 / 6.500664 (-6.377945) 0.061799 / 0.075469 (-0.013670)

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) 1.649178 / 1.841788 (-0.192610) 14.491249 / 8.074308 (6.416940) 26.663143 / 10.191392 (16.471751) 0.935929 / 0.680424 (0.255505) 0.536725 / 0.534201 (0.002524) 0.489136 / 0.579283 (-0.090147) 0.506202 / 0.434364 (0.071838) 0.315357 / 0.540337 (-0.224981) 0.324610 / 1.386936 (-1.062326)

CML watermark

@github-actions
Copy link

Choose a reason for hiding this comment

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

Show benchmarks

PyArrow==3.0.0

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) 0.010278 / 0.011353 (-0.001074) 0.004137 / 0.011008 (-0.006871) 0.034023 / 0.038508 (-0.004485) 0.037324 / 0.023109 (0.014215) 0.349238 / 0.275898 (0.073340) 0.379634 / 0.323480 (0.056154) 0.009076 / 0.007986 (0.001090) 0.005366 / 0.004328 (0.001037) 0.010421 / 0.004250 (0.006171) 0.046155 / 0.037052 (0.009103) 0.346382 / 0.258489 (0.087893) 0.388810 / 0.293841 (0.094969) 0.034033 / 0.128546 (-0.094513) 0.011070 / 0.075646 (-0.064577) 0.295192 / 0.419271 (-0.124080) 0.056653 / 0.043533 (0.013120) 0.337681 / 0.255139 (0.082542) 0.374545 / 0.283200 (0.091346) 0.116106 / 0.141683 (-0.025577) 1.910543 / 1.452155 (0.458388) 2.059256 / 1.492716 (0.566540)

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) 0.254356 / 0.018006 (0.236350) 0.453372 / 0.000490 (0.452882) 0.004197 / 0.000200 (0.003997) 0.000100 / 0.000054 (0.000045)

Benchmark: benchmark_indices_mapping.json

metric select shard shuffle sort train_test_split
new / old (diff) 0.042286 / 0.037411 (0.004875) 0.025347 / 0.014526 (0.010821) 0.032562 / 0.176557 (-0.143994) 0.075013 / 0.737135 (-0.662122) 0.035962 / 0.296338 (-0.260377)

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) 0.452035 / 0.215209 (0.236826) 4.639789 / 2.077655 (2.562135) 2.098367 / 1.504120 (0.594247) 1.752794 / 1.541195 (0.211599) 1.781371 / 1.468490 (0.312880) 0.462930 / 4.584777 (-4.121847) 5.299989 / 3.745712 (1.554277) 4.484025 / 5.269862 (-0.785837) 1.040407 / 4.565676 (-3.525270) 0.062116 / 0.424275 (-0.362159) 0.014671 / 0.007607 (0.007064) 0.573223 / 0.226044 (0.347179) 5.873939 / 2.268929 (3.605011) 2.509747 / 55.444624 (-52.934878) 2.018220 / 6.876477 (-4.858257) 2.049502 / 2.142072 (-0.092570) 0.612571 / 4.805227 (-4.192656) 0.135043 / 6.500664 (-6.365621) 0.067465 / 0.075469 (-0.008004)

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) 1.762780 / 1.841788 (-0.079008) 15.966002 / 8.074308 (7.891694) 30.225630 / 10.191392 (20.034238) 0.909099 / 0.680424 (0.228675) 0.550223 / 0.534201 (0.016022) 0.522708 / 0.579283 (-0.056575) 0.593691 / 0.434364 (0.159327) 0.359422 / 0.540337 (-0.180915) 0.378220 / 1.386936 (-1.008716)
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) 0.008298 / 0.011353 (-0.003055) 0.003955 / 0.011008 (-0.007053) 0.033190 / 0.038508 (-0.005319) 0.036329 / 0.023109 (0.013220) 0.341572 / 0.275898 (0.065674) 0.377845 / 0.323480 (0.054365) 0.006880 / 0.007986 (-0.001106) 0.006942 / 0.004328 (0.002614) 0.008285 / 0.004250 (0.004035) 0.043007 / 0.037052 (0.005955) 0.341660 / 0.258489 (0.083171) 0.390650 / 0.293841 (0.096809) 0.035530 / 0.128546 (-0.093017) 0.010215 / 0.075646 (-0.065431) 0.300349 / 0.419271 (-0.118922) 0.058040 / 0.043533 (0.014507) 0.342847 / 0.255139 (0.087708) 0.383941 / 0.283200 (0.100741) 0.103689 / 0.141683 (-0.037994) 1.919415 / 1.452155 (0.467261) 2.097235 / 1.492716 (0.604519)

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) 0.296958 / 0.018006 (0.278952) 0.452145 / 0.000490 (0.451655) 0.025516 / 0.000200 (0.025316) 0.000435 / 0.000054 (0.000380)

Benchmark: benchmark_indices_mapping.json

metric select shard shuffle sort train_test_split
new / old (diff) 0.038817 / 0.037411 (0.001406) 0.024820 / 0.014526 (0.010294) 0.040711 / 0.176557 (-0.135846) 0.086160 / 0.737135 (-0.650975) 0.037220 / 0.296338 (-0.259118)

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) 0.451069 / 0.215209 (0.235860) 4.577982 / 2.077655 (2.500328) 2.036587 / 1.504120 (0.532467) 1.799284 / 1.541195 (0.258090) 1.861684 / 1.468490 (0.393193) 0.446242 / 4.584777 (-4.138535) 5.366637 / 3.745712 (1.620924) 2.275827 / 5.269862 (-2.994035) 1.037245 / 4.565676 (-3.528432) 0.059063 / 0.424275 (-0.365212) 0.014078 / 0.007607 (0.006471) 0.571040 / 0.226044 (0.344996) 5.804334 / 2.268929 (3.535406) 2.470098 / 55.444624 (-52.974526) 2.074601 / 6.876477 (-4.801876) 2.172418 / 2.142072 (0.030345) 0.609967 / 4.805227 (-4.195260) 0.127493 / 6.500664 (-6.373171) 0.062341 / 0.075469 (-0.013128)

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) 1.762523 / 1.841788 (-0.079265) 15.333798 / 8.074308 (7.259490) 28.746142 / 10.191392 (18.554750) 1.014785 / 0.680424 (0.334361) 0.582172 / 0.534201 (0.047972) 0.562392 / 0.579283 (-0.016891) 0.554860 / 0.434364 (0.120496) 0.345899 / 0.540337 (-0.194438) 0.385531 / 1.386936 (-1.001405)

CML watermark

@github-actions
Copy link

Choose a reason for hiding this comment

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

Show benchmarks

PyArrow==3.0.0

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) 0.011046 / 0.011353 (-0.000307) 0.004583 / 0.011008 (-0.006426) 0.032728 / 0.038508 (-0.005780) 0.040399 / 0.023109 (0.017290) 0.316636 / 0.275898 (0.040738) 0.353190 / 0.323480 (0.029710) 0.009136 / 0.007986 (0.001150) 0.004118 / 0.004328 (-0.000210) 0.009653 / 0.004250 (0.005402) 0.058359 / 0.037052 (0.021307) 0.297285 / 0.258489 (0.038796) 0.358107 / 0.293841 (0.064266) 0.032924 / 0.128546 (-0.095622) 0.010298 / 0.075646 (-0.065348) 0.260878 / 0.419271 (-0.158393) 0.052845 / 0.043533 (0.009312) 0.299455 / 0.255139 (0.044316) 0.321208 / 0.283200 (0.038008) 0.111410 / 0.141683 (-0.030273) 1.804073 / 1.452155 (0.351919) 1.868163 / 1.492716 (0.375447)

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) 0.318303 / 0.018006 (0.300297) 0.619390 / 0.000490 (0.618900) 0.007448 / 0.000200 (0.007248) 0.000117 / 0.000054 (0.000063)

Benchmark: benchmark_indices_mapping.json

metric select shard shuffle sort train_test_split
new / old (diff) 0.038292 / 0.037411 (0.000881) 0.024407 / 0.014526 (0.009881) 0.033772 / 0.176557 (-0.142784) 0.076364 / 0.737135 (-0.660772) 0.036437 / 0.296338 (-0.259901)

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) 0.430819 / 0.215209 (0.215610) 4.299455 / 2.077655 (2.221801) 1.849588 / 1.504120 (0.345468) 1.642642 / 1.541195 (0.101447) 1.774262 / 1.468490 (0.305771) 0.449754 / 4.584777 (-4.135023) 4.644184 / 3.745712 (0.898472) 2.362596 / 5.269862 (-2.907266) 0.943201 / 4.565676 (-3.622475) 0.054232 / 0.424275 (-0.370043) 0.012463 / 0.007607 (0.004856) 0.532433 / 0.226044 (0.306388) 5.367544 / 2.268929 (3.098615) 2.330514 / 55.444624 (-53.114110) 1.944966 / 6.876477 (-4.931511) 2.045770 / 2.142072 (-0.096302) 0.569599 / 4.805227 (-4.235629) 0.126478 / 6.500664 (-6.374186) 0.064001 / 0.075469 (-0.011468)

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) 1.614755 / 1.841788 (-0.227032) 15.655681 / 8.074308 (7.581373) 27.748033 / 10.191392 (17.556641) 0.879977 / 0.680424 (0.199554) 0.525630 / 0.534201 (-0.008571) 0.501213 / 0.579283 (-0.078070) 0.518386 / 0.434364 (0.084022) 0.328722 / 0.540337 (-0.211615) 0.344547 / 1.386936 (-1.042389)
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) 0.008935 / 0.011353 (-0.002418) 0.004349 / 0.011008 (-0.006660) 0.030082 / 0.038508 (-0.008426) 0.036726 / 0.023109 (0.013617) 0.308179 / 0.275898 (0.032281) 0.336494 / 0.323480 (0.013014) 0.006722 / 0.007986 (-0.001264) 0.005296 / 0.004328 (0.000967) 0.007680 / 0.004250 (0.003430) 0.044669 / 0.037052 (0.007617) 0.295083 / 0.258489 (0.036594) 0.336568 / 0.293841 (0.042727) 0.032312 / 0.128546 (-0.096235) 0.009947 / 0.075646 (-0.065699) 0.253040 / 0.419271 (-0.166231) 0.051663 / 0.043533 (0.008130) 0.300023 / 0.255139 (0.044884) 0.328969 / 0.283200 (0.045769) 0.099921 / 0.141683 (-0.041761) 1.762552 / 1.452155 (0.310397) 1.823781 / 1.492716 (0.331064)

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) 0.344459 / 0.018006 (0.326453) 0.606013 / 0.000490 (0.605524) 0.008420 / 0.000200 (0.008220) 0.000249 / 0.000054 (0.000194)

Benchmark: benchmark_indices_mapping.json

metric select shard shuffle sort train_test_split
new / old (diff) 0.033805 / 0.037411 (-0.003607) 0.023453 / 0.014526 (0.008928) 0.034329 / 0.176557 (-0.142228) 0.078573 / 0.737135 (-0.658562) 0.035978 / 0.296338 (-0.260360)

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) 0.417249 / 0.215209 (0.202040) 4.183013 / 2.077655 (2.105358) 1.786974 / 1.504120 (0.282855) 1.586537 / 1.541195 (0.045342) 1.724462 / 1.468490 (0.255972) 0.442130 / 4.584777 (-4.142647) 4.680934 / 3.745712 (0.935222) 3.644956 / 5.269862 (-1.624906) 0.963763 / 4.565676 (-3.601914) 0.054140 / 0.424275 (-0.370135) 0.012582 / 0.007607 (0.004975) 0.525566 / 0.226044 (0.299521) 5.243712 / 2.268929 (2.974783) 2.285815 / 55.444624 (-53.158810) 1.911536 / 6.876477 (-4.964940) 2.111712 / 2.142072 (-0.030360) 0.570082 / 4.805227 (-4.235145) 0.126242 / 6.500664 (-6.374422) 0.064453 / 0.075469 (-0.011016)

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) 1.679198 / 1.841788 (-0.162590) 15.482951 / 8.074308 (7.408643) 28.180963 / 10.191392 (17.989571) 0.906580 / 0.680424 (0.226156) 0.566406 / 0.534201 (0.032205) 0.528301 / 0.579283 (-0.050982) 0.515327 / 0.434364 (0.080963) 0.342609 / 0.540337 (-0.197729) 0.355470 / 1.386936 (-1.031466)

CML watermark

Please sign in to comment.