Skip to content

Commit

Permalink
update dataset_infos.json
Browse files Browse the repository at this point in the history
  • Loading branch information
@lhoestq
lhoestq committed Mar 24, 2022
1 parent ed39389 commit 27fa5ef
Show file tree
Hide file tree
Showing 6 changed files with 6 additions and 6 deletions.
2 changes: 1 addition & 1 deletion datasets/ami/dataset_infos.json
View file

Large diffs are not rendered by default.

2 changes: 1 addition & 1 deletion datasets/arabic_speech_corpus/dataset_infos.json
View file
Original file line number Diff line number Diff line change
@@ -1 +1 @@
{"clean": {"description": "This Speech corpus has been developed as part of PhD work carried out by Nawar Halabi at the University of Southampton.\nThe corpus was recorded in south Levantine Arabic\n(Damascian accent) using a professional studio. Synthesized speech as an output using this corpus has produced a high quality, natural voice.\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .flac format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport soundfile as sf\n\ndef map_to_array(batch):\n speech_array, _ = sf.read(batch[\"file\"])\n batch[\"speech\"] = speech_array\n return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "@phdthesis{halabi2016modern,\n title={Modern standard Arabic phonetics for speech synthesis},\n author={Halabi, Nawar},\n year={2016},\n school={University of Southampton}\n}\n", "homepage": "http://en.arabicspeechcorpus.com/arabic-speech-corpus.zip", "license": "", "features": {"file": {"dtype": "string", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}, "phonetic": {"dtype": "string", "id": null, "_type": "Value"}, "orthographic": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "text"}, "task_templates": [{"task": "automatic-speech-recognition", "audio_column": "audio", "transcription_column": "text"}], "builder_name": "arabic_speech_corpus", "config_name": "clean", "version": {"version_str": "2.1.0", "description": "", "major": 2, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 1002365, "num_examples": 1813, "dataset_name": "arabic_speech_corpus"}, "test": {"name": "test", "num_bytes": 65784, "num_examples": 100, "dataset_name": "arabic_speech_corpus"}}, "download_checksums": {"http://en.arabicspeechcorpus.com/arabic-speech-corpus.zip": {"num_bytes": 1192302846, "checksum": "1df85219370fb1ebe8bfc46aa886265586411d04e7c1caa5a5b9847b3ad5f9de"}}, "download_size": 1192302846, "post_processing_size": null, "dataset_size": 1068149, "size_in_bytes": 1193370995}}
{"clean": {"description": "This Speech corpus has been developed as part of PhD work carried out by Nawar Halabi at the University of Southampton.\nThe corpus was recorded in south Levantine Arabic\n(Damascian accent) using a professional studio. Synthesized speech as an output using this corpus has produced a high quality, natural voice.\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .flac format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport soundfile as sf\n\ndef map_to_array(batch):\n speech_array, _ = sf.read(batch[\"file\"])\n batch[\"speech\"] = speech_array\n return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "@phdthesis{halabi2016modern,\n title={Modern standard Arabic phonetics for speech synthesis},\n author={Halabi, Nawar},\n year={2016},\n school={University of Southampton}\n}\n", "homepage": "http://en.arabicspeechcorpus.com/arabic-speech-corpus.zip", "license": "", "features": {"file": {"dtype": "string", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}, "audio": {"sampling_rate": 48000, "mono": true, "decode": true, "id": null, "_type": "Audio"}, "phonetic": {"dtype": "string", "id": null, "_type": "Value"}, "orthographic": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "text"}, "task_templates": [{"task": "automatic-speech-recognition", "audio_column": "audio", "transcription_column": "text"}], "builder_name": "arabic_speech_corpus", "config_name": "clean", "version": {"version_str": "2.1.0", "description": "", "major": 2, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 1002365, "num_examples": 1813, "dataset_name": "arabic_speech_corpus"}, "test": {"name": "test", "num_bytes": 65784, "num_examples": 100, "dataset_name": "arabic_speech_corpus"}}, "download_checksums": {"http://en.arabicspeechcorpus.com/arabic-speech-corpus.zip": {"num_bytes": 1192302846, "checksum": "1df85219370fb1ebe8bfc46aa886265586411d04e7c1caa5a5b9847b3ad5f9de"}}, "download_size": 1192302846, "post_processing_size": null, "dataset_size": 1068149, "size_in_bytes": 1193370995}}
2 changes: 1 addition & 1 deletion datasets/lj_speech/dataset_infos.json
View file
Original file line number Diff line number Diff line change
@@ -1 +1 @@
{"main": {"description": "This is a public domain speech dataset consisting of 13,100 short audio clips of a single speaker reading \npassages from 7 non-fiction books in English. A transcription is provided for each clip. Clips vary in length \nfrom 1 to 10 seconds and have a total length of approximately 24 hours.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .wav format and is not converted to a float32 array. To convert the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport soundfile as sf\n\ndef map_to_array(batch):\n speech_array, _ = sf.read(batch[\"file\"])\n batch[\"speech\"] = speech_array\n return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "@misc{ljspeech17,\n author = {Keith Ito and Linda Johnson},\n title = {The LJ Speech Dataset},\n howpublished = {\\url{https://keithito.com/LJ-Speech-Dataset/}},\n year = 2017\n}\n", "homepage": "https://keithito.com/LJ-Speech-Dataset/", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}, "normalized_text": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "text"}, "task_templates": [{"task": "automatic-speech-recognition", "audio_column": "audio", "transcription_column": "text"}], "builder_name": "lj_speech", "config_name": "main", "version": {"version_str": "1.1.0", "description": null, "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 4667022, "num_examples": 13100, "dataset_name": "lj_speech"}}, "download_checksums": {"https://data.keithito.com/data/speech/LJSpeech-1.1.tar.bz2": {"num_bytes": 2748572632, "checksum": "be1a30453f28eb8dd26af4101ae40cbf2c50413b1bb21936cbcdc6fae3de8aa5"}}, "download_size": 2748572632, "post_processing_size": null, "dataset_size": 4667022, "size_in_bytes": 2753239654}}
{"main": {"description": "This is a public domain speech dataset consisting of 13,100 short audio clips of a single speaker reading \npassages from 7 non-fiction books in English. A transcription is provided for each clip. Clips vary in length \nfrom 1 to 10 seconds and have a total length of approximately 24 hours.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .wav format and is not converted to a float32 array. To convert the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport soundfile as sf\n\ndef map_to_array(batch):\n speech_array, _ = sf.read(batch[\"file\"])\n batch[\"speech\"] = speech_array\n return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "@misc{ljspeech17,\n author = {Keith Ito and Linda Johnson},\n title = {The LJ Speech Dataset},\n howpublished = {\\url{https://keithito.com/LJ-Speech-Dataset/}},\n year = 2017\n}\n", "homepage": "https://keithito.com/LJ-Speech-Dataset/", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "audio": {"sampling_rate": 22050, "mono": true, "decode": true, "id": null, "_type": "Audio"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}, "normalized_text": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "text"}, "task_templates": [{"task": "automatic-speech-recognition", "audio_column": "audio", "transcription_column": "text"}], "builder_name": "lj_speech", "config_name": "main", "version": {"version_str": "1.1.0", "description": null, "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 4667022, "num_examples": 13100, "dataset_name": "lj_speech"}}, "download_checksums": {"https://data.keithito.com/data/speech/LJSpeech-1.1.tar.bz2": {"num_bytes": 2748572632, "checksum": "be1a30453f28eb8dd26af4101ae40cbf2c50413b1bb21936cbcdc6fae3de8aa5"}}, "download_size": 2748572632, "post_processing_size": null, "dataset_size": 4667022, "size_in_bytes": 2753239654}}
2 changes: 1 addition & 1 deletion datasets/superb/dataset_infos.json
View file

Large diffs are not rendered by default.

2 changes: 1 addition & 1 deletion datasets/superb/superb.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -147,7 +147,7 @@ class Superb(datasets.GeneratorBasedBuilder):
supervised_keys=("file", "text"),
url="http://www.openslr.org/12",
data_url="http://www.openslr.org/resources/12/",
task_templates=[AutomaticSpeechRecognition(audio="audio_column", transcription_column="text")],
task_templates=[AutomaticSpeechRecognition(audio_column="audio", transcription_column="text")],
),
SuperbConfig(
name="ks",
Expand Down
2 changes: 1 addition & 1 deletion datasets/timit_asr/dataset_infos.json
View file
Original file line number Diff line number Diff line change
@@ -1 +1 @@
{"clean": {"description": "The TIMIT corpus of reading speech has been developed to provide speech data for acoustic-phonetic research studies\nand for the evaluation of automatic speech recognition systems.\n\nTIMIT contains high quality recordings of 630 individuals/speakers with 8 different American English dialects,\nwith each individual reading upto 10 phonetically rich sentences.\n\nMore info on TIMIT dataset can be understood from the \"README\" which can be found here:\nhttps://catalog.ldc.upenn.edu/docs/LDC93S1/readme.txt\n", "citation": "@inproceedings{\n title={TIMIT Acoustic-Phonetic Continuous Speech Corpus},\n author={Garofolo, John S., et al},\n ldc_catalog_no={LDC93S1},\n DOI={https://doi.org/10.35111/17gk-bn40},\n journal={Linguistic Data Consortium, Philadelphia},\n year={1983}\n}\n", "homepage": "https://catalog.ldc.upenn.edu/LDC93S1", "license": "", "features": {"file": {"dtype": "string", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}, "phonetic_detail": {"feature": {"start": {"dtype": "int64", "id": null, "_type": "Value"}, "stop": {"dtype": "int64", "id": null, "_type": "Value"}, "utterance": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}, "word_detail": {"feature": {"start": {"dtype": "int64", "id": null, "_type": "Value"}, "stop": {"dtype": "int64", "id": null, "_type": "Value"}, "utterance": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}, "dialect_region": {"dtype": "string", "id": null, "_type": "Value"}, "sentence_type": {"dtype": "string", "id": null, "_type": "Value"}, "speaker_id": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "text"}, "task_templates": [{"task": "automatic-speech-recognition", "audio_column": "audio", "transcription_column": "text"}], "builder_name": "timit_asr", "config_name": "clean", "version": {"version_str": "2.0.1", "description": "", "major": 2, "minor": 0, "patch": 1}, "splits": {"train": {"name": "train", "num_bytes": 6076580, "num_examples": 4620, "dataset_name": "timit_asr"}, "test": {"name": "test", "num_bytes": 2202968, "num_examples": 1680, "dataset_name": "timit_asr"}}, "download_checksums": {"https://data.deepai.org/timit.zip": {"num_bytes": 869007403, "checksum": "b79af42068b53045510d86854e2239a13ff77c4bd27803b40c28dce4bb5aeb0d"}}, "download_size": 869007403, "post_processing_size": null, "dataset_size": 8279548, "size_in_bytes": 877286951}}
{"clean": {"description": "The TIMIT corpus of reading speech has been developed to provide speech data for acoustic-phonetic research studies\nand for the evaluation of automatic speech recognition systems.\n\nTIMIT contains high quality recordings of 630 individuals/speakers with 8 different American English dialects,\nwith each individual reading upto 10 phonetically rich sentences.\n\nMore info on TIMIT dataset can be understood from the \"README\" which can be found here:\nhttps://catalog.ldc.upenn.edu/docs/LDC93S1/readme.txt\n", "citation": "@inproceedings{\n title={TIMIT Acoustic-Phonetic Continuous Speech Corpus},\n author={Garofolo, John S., et al},\n ldc_catalog_no={LDC93S1},\n DOI={https://doi.org/10.35111/17gk-bn40},\n journal={Linguistic Data Consortium, Philadelphia},\n year={1983}\n}\n", "homepage": "https://catalog.ldc.upenn.edu/LDC93S1", "license": "", "features": {"file": {"dtype": "string", "id": null, "_type": "Value"}, "audio": {"sampling_rate": 16000, "mono": true, "decode": true, "id": null, "_type": "Audio"}, "text": {"dtype": "string", "id": null, "_type": "Value"}, "phonetic_detail": {"feature": {"start": {"dtype": "int64", "id": null, "_type": "Value"}, "stop": {"dtype": "int64", "id": null, "_type": "Value"}, "utterance": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}, "word_detail": {"feature": {"start": {"dtype": "int64", "id": null, "_type": "Value"}, "stop": {"dtype": "int64", "id": null, "_type": "Value"}, "utterance": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}, "dialect_region": {"dtype": "string", "id": null, "_type": "Value"}, "sentence_type": {"dtype": "string", "id": null, "_type": "Value"}, "speaker_id": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "text"}, "task_templates": [{"task": "automatic-speech-recognition", "audio_column": "audio", "transcription_column": "text"}], "builder_name": "timit_asr", "config_name": "clean", "version": {"version_str": "2.0.1", "description": "", "major": 2, "minor": 0, "patch": 1}, "splits": {"train": {"name": "train", "num_bytes": 6076580, "num_examples": 4620, "dataset_name": "timit_asr"}, "test": {"name": "test", "num_bytes": 2202968, "num_examples": 1680, "dataset_name": "timit_asr"}}, "download_checksums": {"https://data.deepai.org/timit.zip": {"num_bytes": 869007403, "checksum": "b79af42068b53045510d86854e2239a13ff77c4bd27803b40c28dce4bb5aeb0d"}}, "download_size": 869007403, "post_processing_size": null, "dataset_size": 8279548, "size_in_bytes": 877286951}}

1 comment on commit 27fa5ef

@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==5.0.0

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) 0.011717 / 0.011353 (0.000364) 0.004560 / 0.011008 (-0.006448) 0.036596 / 0.038508 (-0.001912) 0.042104 / 0.023109 (0.018995) 0.360865 / 0.275898 (0.084967) 0.371586 / 0.323480 (0.048106) 0.009277 / 0.007986 (0.001292) 0.003946 / 0.004328 (-0.000383) 0.010254 / 0.004250 (0.006004) 0.047133 / 0.037052 (0.010081) 0.335247 / 0.258489 (0.076758) 0.369114 / 0.293841 (0.075273) 0.037401 / 0.128546 (-0.091146) 0.011427 / 0.075646 (-0.064219) 0.301917 / 0.419271 (-0.117354) 0.058767 / 0.043533 (0.015234) 0.328711 / 0.255139 (0.073572) 0.366183 / 0.283200 (0.082984) 0.120090 / 0.141683 (-0.021592) 2.059229 / 1.452155 (0.607074) 2.108362 / 1.492716 (0.615646)

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) 0.346413 / 0.018006 (0.328407) 0.474576 / 0.000490 (0.474087) 0.055094 / 0.000200 (0.054894) 0.000688 / 0.000054 (0.000634)

Benchmark: benchmark_indices_mapping.json

metric select shard shuffle sort train_test_split
new / old (diff) 0.032690 / 0.037411 (-0.004722) 0.130407 / 0.014526 (0.115882) 0.133167 / 0.176557 (-0.043390) 0.177712 / 0.737135 (-0.559423) 0.136912 / 0.296338 (-0.159427)

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) 0.485860 / 0.215209 (0.270651) 4.897182 / 2.077655 (2.819528) 2.100872 / 1.504120 (0.596752) 1.859856 / 1.541195 (0.318662) 1.897317 / 1.468490 (0.428827) 0.520519 / 4.584777 (-4.064258) 5.574757 / 3.745712 (1.829045) 2.547897 / 5.269862 (-2.721964) 1.078347 / 4.565676 (-3.487330) 0.063155 / 0.424275 (-0.361120) 0.014835 / 0.007607 (0.007228) 0.601770 / 0.226044 (0.375725) 6.055309 / 2.268929 (3.786380) 2.630866 / 55.444624 (-52.813759) 2.165453 / 6.876477 (-4.711024) 2.282045 / 2.142072 (0.139973) 0.645869 / 4.805227 (-4.159358) 0.144669 / 6.500664 (-6.355995) 0.070674 / 0.075469 (-0.004795)

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) 1.921933 / 1.841788 (0.080145) 16.440658 / 8.074308 (8.366350) 31.227352 / 10.191392 (21.035960) 1.009870 / 0.680424 (0.329446) 0.623530 / 0.534201 (0.089329) 0.575407 / 0.579283 (-0.003876) 0.628287 / 0.434364 (0.193923) 0.380192 / 0.540337 (-0.160145) 0.408861 / 1.386936 (-0.978075)
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) 0.009570 / 0.011353 (-0.001783) 0.004189 / 0.011008 (-0.006819) 0.035357 / 0.038508 (-0.003151) 0.038548 / 0.023109 (0.015439) 0.389192 / 0.275898 (0.113294) 0.396917 / 0.323480 (0.073437) 0.006975 / 0.007986 (-0.001011) 0.005486 / 0.004328 (0.001157) 0.008652 / 0.004250 (0.004402) 0.043614 / 0.037052 (0.006562) 0.362723 / 0.258489 (0.104234) 0.399530 / 0.293841 (0.105689) 0.035210 / 0.128546 (-0.093337) 0.011193 / 0.075646 (-0.064454) 0.310754 / 0.419271 (-0.108518) 0.056989 / 0.043533 (0.013456) 0.397758 / 0.255139 (0.142619) 0.407610 / 0.283200 (0.124410) 0.101421 / 0.141683 (-0.040262) 2.026184 / 1.452155 (0.574029) 2.125905 / 1.492716 (0.633188)

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) 0.310601 / 0.018006 (0.292595) 0.477751 / 0.000490 (0.477261) 0.020529 / 0.000200 (0.020329) 0.000517 / 0.000054 (0.000463)

Benchmark: benchmark_indices_mapping.json

metric select shard shuffle sort train_test_split
new / old (diff) 0.029053 / 0.037411 (-0.008359) 0.122348 / 0.014526 (0.107822) 0.131390 / 0.176557 (-0.045166) 0.188579 / 0.737135 (-0.548556) 0.134976 / 0.296338 (-0.161362)

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) 0.492892 / 0.215209 (0.277683) 4.946025 / 2.077655 (2.868371) 2.147864 / 1.504120 (0.643744) 1.908723 / 1.541195 (0.367529) 1.974644 / 1.468490 (0.506154) 0.508285 / 4.584777 (-4.076491) 5.557338 / 3.745712 (1.811625) 2.452268 / 5.269862 (-2.817594) 1.091136 / 4.565676 (-3.474541) 0.062607 / 0.424275 (-0.361668) 0.014578 / 0.007607 (0.006971) 0.614085 / 0.226044 (0.388040) 6.087146 / 2.268929 (3.818218) 2.635586 / 55.444624 (-52.809038) 2.212791 / 6.876477 (-4.663685) 2.299611 / 2.142072 (0.157538) 0.643121 / 4.805227 (-4.162106) 0.141621 / 6.500664 (-6.359043) 0.067978 / 0.075469 (-0.007491)

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) 1.896066 / 1.841788 (0.054278) 15.880933 / 8.074308 (7.806625) 31.056633 / 10.191392 (20.865241) 1.005944 / 0.680424 (0.325520) 0.612275 / 0.534201 (0.078074) 0.583915 / 0.579283 (0.004631) 0.628135 / 0.434364 (0.193771) 0.382537 / 0.540337 (-0.157801) 0.392092 / 1.386936 (-0.994844)

CML watermark

Please sign in to comment.