albertvillanova HF staff commited on
Commit
45d4cbe
1 Parent(s): 320355e

Convert dataset to Parquet

Browse files

Convert dataset to Parquet.

README.md CHANGED
@@ -26,16 +26,16 @@ dataset_info:
26
  sequence: string
27
  splits:
28
  - name: validation
29
- num_bytes: 42710107
30
  num_examples: 10047
31
  - name: train
32
- num_bytes: 350884446
33
  num_examples: 82326
34
  - name: test
35
- num_bytes: 41020711
36
  num_examples: 9650
37
- download_size: 168698008
38
- dataset_size: 434615264
39
  - config_name: v2.1
40
  features:
41
  - name: answers
@@ -68,6 +68,15 @@ dataset_info:
68
  num_examples: 101092
69
  download_size: 1384271865
70
  dataset_size: 4287455242
 
 
 
 
 
 
 
 
 
71
  ---
72
 
73
  # Dataset Card for "ms_marco"
 
26
  sequence: string
27
  splits:
28
  - name: validation
29
+ num_bytes: 42665198
30
  num_examples: 10047
31
  - name: train
32
+ num_bytes: 350516260
33
  num_examples: 82326
34
  - name: test
35
+ num_bytes: 40977580
36
  num_examples: 9650
37
+ download_size: 217328153
38
+ dataset_size: 434159038
39
  - config_name: v2.1
40
  features:
41
  - name: answers
 
68
  num_examples: 101092
69
  download_size: 1384271865
70
  dataset_size: 4287455242
71
+ configs:
72
+ - config_name: v1.1
73
+ data_files:
74
+ - split: validation
75
+ path: v1.1/validation-*
76
+ - split: train
77
+ path: v1.1/train-*
78
+ - split: test
79
+ path: v1.1/test-*
80
  ---
81
 
82
  # Dataset Card for "ms_marco"
dataset_infos.json CHANGED
@@ -1 +1,199 @@
1
- {"v1.1": {"description": "\nStarting with a paper released at NIPS 2016, MS MARCO is a collection of datasets focused on deep learning in search.\n\nThe first dataset was a question answering dataset featuring 100,000 real Bing questions and a human generated answer. \nSince then we released a 1,000,000 question dataset, a natural langauge generation dataset, a passage ranking dataset, \nkeyphrase extraction dataset, crawling dataset, and a conversational search.\n\nThere have been 277 submissions. 20 KeyPhrase Extraction submissions, 87 passage ranking submissions, 0 document ranking \nsubmissions, 73 QnA V2 submissions, 82 NLGEN submisions, and 15 QnA V1 submissions\n\nThis data comes in three tasks/forms: Original QnA dataset(v1.1), Question Answering(v2.1), Natural Language Generation(v2.1). \n\nThe original question answering datset featured 100,000 examples and was released in 2016. Leaderboard is now closed but data is availible below.\n\nThe current competitive tasks are Question Answering and Natural Language Generation. Question Answering features over 1,000,000 queries and \nis much like the original QnA dataset but bigger and with higher quality. The Natural Language Generation dataset features 180,000 examples and \nbuilds upon the QnA dataset to deliver answers that could be spoken by a smart speaker.\n\n\nversion v1.1", "citation": "\n@article{DBLP:journals/corr/NguyenRSGTMD16,\n author = {Tri Nguyen and\n Mir Rosenberg and\n Xia Song and\n Jianfeng Gao and\n Saurabh Tiwary and\n Rangan Majumder and\n Li Deng},\n title = {{MS} {MARCO:} {A} Human Generated MAchine Reading COmprehension Dataset},\n journal = {CoRR},\n volume = {abs/1611.09268},\n year = {2016},\n url = {http://arxiv.org/abs/1611.09268},\n archivePrefix = {arXiv},\n eprint = {1611.09268},\n timestamp = {Mon, 13 Aug 2018 16:49:03 +0200},\n biburl = {https://dblp.org/rec/journals/corr/NguyenRSGTMD16.bib},\n bibsource = {dblp computer science bibliography, https://dblp.org}\n}\n}\n", "homepage": "https://microsoft.github.io/msmarco/", "license": "", "features": {"answers": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "passages": {"feature": {"is_selected": {"dtype": "int32", "id": null, "_type": "Value"}, "passage_text": {"dtype": "string", "id": null, "_type": "Value"}, "url": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}, "query": {"dtype": "string", "id": null, "_type": "Value"}, "query_id": {"dtype": "int32", "id": null, "_type": "Value"}, "query_type": {"dtype": "string", "id": null, "_type": "Value"}, "wellFormedAnswers": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "supervised_keys": null, "builder_name": "ms_marco", "config_name": "v1.1", "version": {"version_str": "1.1.0", "description": "", "datasets_version_to_prepare": null, "major": 1, "minor": 1, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 42710107, "num_examples": 10047, "dataset_name": "ms_marco"}, "train": {"name": "train", "num_bytes": 350884446, "num_examples": 82326, "dataset_name": "ms_marco"}, "test": {"name": "test", "num_bytes": 41020711, "num_examples": 9650, "dataset_name": "ms_marco"}}, "download_checksums": {"https://msmarco.blob.core.windows.net/msmsarcov1/train_v1.1.json.gz": {"num_bytes": 110704491, "checksum": "2aaa60df3a758137f0bb7c01fe334858477eb46fa8665ea01588e553cda6aa9f"}, "https://msmarco.blob.core.windows.net/msmsarcov1/dev_v1.1.json.gz": {"num_bytes": 13493661, "checksum": "c70fcb1de78e635cf501264891a1a56d52e7f63e69623da7dd41d89a785d67ca"}, "https://msmarco.blob.core.windows.net/msmsarcov1/test_hidden_v1.1.json": {"num_bytes": 44499856, "checksum": "083aa4f4d86ba0cedb830ca9972eff69f73cbc32b1da26b8617205f0dedea757"}}, "download_size": 168698008, "dataset_size": 434615264, "size_in_bytes": 603313272}, "v2.1": {"description": "\nStarting with a paper released at NIPS 2016, MS MARCO is a collection of datasets focused on deep learning in search.\n\nThe first dataset was a question answering dataset featuring 100,000 real Bing questions and a human generated answer. \nSince then we released a 1,000,000 question dataset, a natural langauge generation dataset, a passage ranking dataset, \nkeyphrase extraction dataset, crawling dataset, and a conversational search.\n\nThere have been 277 submissions. 20 KeyPhrase Extraction submissions, 87 passage ranking submissions, 0 document ranking \nsubmissions, 73 QnA V2 submissions, 82 NLGEN submisions, and 15 QnA V1 submissions\n\nThis data comes in three tasks/forms: Original QnA dataset(v1.1), Question Answering(v2.1), Natural Language Generation(v2.1). \n\nThe original question answering datset featured 100,000 examples and was released in 2016. Leaderboard is now closed but data is availible below.\n\nThe current competitive tasks are Question Answering and Natural Language Generation. Question Answering features over 1,000,000 queries and \nis much like the original QnA dataset but bigger and with higher quality. The Natural Language Generation dataset features 180,000 examples and \nbuilds upon the QnA dataset to deliver answers that could be spoken by a smart speaker.\n\n\nversion v2.1", "citation": "\n@article{DBLP:journals/corr/NguyenRSGTMD16,\n author = {Tri Nguyen and\n Mir Rosenberg and\n Xia Song and\n Jianfeng Gao and\n Saurabh Tiwary and\n Rangan Majumder and\n Li Deng},\n title = {{MS} {MARCO:} {A} Human Generated MAchine Reading COmprehension Dataset},\n journal = {CoRR},\n volume = {abs/1611.09268},\n year = {2016},\n url = {http://arxiv.org/abs/1611.09268},\n archivePrefix = {arXiv},\n eprint = {1611.09268},\n timestamp = {Mon, 13 Aug 2018 16:49:03 +0200},\n biburl = {https://dblp.org/rec/journals/corr/NguyenRSGTMD16.bib},\n bibsource = {dblp computer science bibliography, https://dblp.org}\n}\n}\n", "homepage": "https://microsoft.github.io/msmarco/", "license": "", "features": {"answers": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "passages": {"feature": {"is_selected": {"dtype": "int32", "id": null, "_type": "Value"}, "passage_text": {"dtype": "string", "id": null, "_type": "Value"}, "url": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}, "query": {"dtype": "string", "id": null, "_type": "Value"}, "query_id": {"dtype": "int32", "id": null, "_type": "Value"}, "query_type": {"dtype": "string", "id": null, "_type": "Value"}, "wellFormedAnswers": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "supervised_keys": null, "builder_name": "ms_marco", "config_name": "v2.1", "version": {"version_str": "2.1.0", "description": "", "datasets_version_to_prepare": null, "major": 2, "minor": 1, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 414286005, "num_examples": 101093, "dataset_name": "ms_marco"}, "train": {"name": "train", "num_bytes": 3466972085, "num_examples": 808731, "dataset_name": "ms_marco"}, "test": {"name": "test", "num_bytes": 406197152, "num_examples": 101092, "dataset_name": "ms_marco"}}, "download_checksums": {"https://msmarco.blob.core.windows.net/msmarco/train_v2.1.json.gz": {"num_bytes": 1112116929, "checksum": "e91745411ca81e441a3bb75deb71ce000dc2fc31334085b7d499982f14218fe2"}, "https://msmarco.blob.core.windows.net/msmarco/dev_v2.1.json.gz": {"num_bytes": 138303699, "checksum": "5b3c9c20d1808ee199a930941b0d96f79e397e9234f77a1496890b138df7cb3c"}, "https://msmarco.blob.core.windows.net/msmarco/eval_v2.1_public.json.gz": {"num_bytes": 133851237, "checksum": "05ac0e448450d507e7ff8e37f48a41cc2d015f5bd2c7974d2445f00a53625db6"}}, "download_size": 1384271865, "dataset_size": 4287455242, "size_in_bytes": 5671727107}}
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "v1.1": {
3
+ "description": "\nStarting with a paper released at NIPS 2016, MS MARCO is a collection of datasets focused on deep learning in search.\n\nThe first dataset was a question answering dataset featuring 100,000 real Bing questions and a human generated answer.\nSince then we released a 1,000,000 question dataset, a natural langauge generation dataset, a passage ranking dataset,\nkeyphrase extraction dataset, crawling dataset, and a conversational search.\n\nThere have been 277 submissions. 20 KeyPhrase Extraction submissions, 87 passage ranking submissions, 0 document ranking\nsubmissions, 73 QnA V2 submissions, 82 NLGEN submisions, and 15 QnA V1 submissions\n\nThis data comes in three tasks/forms: Original QnA dataset(v1.1), Question Answering(v2.1), Natural Language Generation(v2.1).\n\nThe original question answering datset featured 100,000 examples and was released in 2016. Leaderboard is now closed but data is availible below.\n\nThe current competitive tasks are Question Answering and Natural Language Generation. Question Answering features over 1,000,000 queries and\nis much like the original QnA dataset but bigger and with higher quality. The Natural Language Generation dataset features 180,000 examples and\nbuilds upon the QnA dataset to deliver answers that could be spoken by a smart speaker.\n\n\nversion v1.1",
4
+ "citation": "\n@article{DBLP:journals/corr/NguyenRSGTMD16,\n author = {Tri Nguyen and\n Mir Rosenberg and\n Xia Song and\n Jianfeng Gao and\n Saurabh Tiwary and\n Rangan Majumder and\n Li Deng},\n title = {{MS} {MARCO:} {A} Human Generated MAchine Reading COmprehension Dataset},\n journal = {CoRR},\n volume = {abs/1611.09268},\n year = {2016},\n url = {http://arxiv.org/abs/1611.09268},\n archivePrefix = {arXiv},\n eprint = {1611.09268},\n timestamp = {Mon, 13 Aug 2018 16:49:03 +0200},\n biburl = {https://dblp.org/rec/journals/corr/NguyenRSGTMD16.bib},\n bibsource = {dblp computer science bibliography, https://dblp.org}\n}\n}\n",
5
+ "homepage": "https://microsoft.github.io/msmarco/",
6
+ "license": "",
7
+ "features": {
8
+ "answers": {
9
+ "feature": {
10
+ "dtype": "string",
11
+ "_type": "Value"
12
+ },
13
+ "_type": "Sequence"
14
+ },
15
+ "passages": {
16
+ "feature": {
17
+ "is_selected": {
18
+ "dtype": "int32",
19
+ "_type": "Value"
20
+ },
21
+ "passage_text": {
22
+ "dtype": "string",
23
+ "_type": "Value"
24
+ },
25
+ "url": {
26
+ "dtype": "string",
27
+ "_type": "Value"
28
+ }
29
+ },
30
+ "_type": "Sequence"
31
+ },
32
+ "query": {
33
+ "dtype": "string",
34
+ "_type": "Value"
35
+ },
36
+ "query_id": {
37
+ "dtype": "int32",
38
+ "_type": "Value"
39
+ },
40
+ "query_type": {
41
+ "dtype": "string",
42
+ "_type": "Value"
43
+ },
44
+ "wellFormedAnswers": {
45
+ "feature": {
46
+ "dtype": "string",
47
+ "_type": "Value"
48
+ },
49
+ "_type": "Sequence"
50
+ }
51
+ },
52
+ "builder_name": "ms_marco",
53
+ "dataset_name": "ms_marco",
54
+ "config_name": "v1.1",
55
+ "version": {
56
+ "version_str": "1.1.0",
57
+ "description": "",
58
+ "major": 1,
59
+ "minor": 1,
60
+ "patch": 0
61
+ },
62
+ "splits": {
63
+ "validation": {
64
+ "name": "validation",
65
+ "num_bytes": 42665198,
66
+ "num_examples": 10047,
67
+ "dataset_name": null
68
+ },
69
+ "train": {
70
+ "name": "train",
71
+ "num_bytes": 350516260,
72
+ "num_examples": 82326,
73
+ "dataset_name": null
74
+ },
75
+ "test": {
76
+ "name": "test",
77
+ "num_bytes": 40977580,
78
+ "num_examples": 9650,
79
+ "dataset_name": null
80
+ }
81
+ },
82
+ "download_size": 217328153,
83
+ "dataset_size": 434159038,
84
+ "size_in_bytes": 651487191
85
+ },
86
+ "v2.1": {
87
+ "description": "\nStarting with a paper released at NIPS 2016, MS MARCO is a collection of datasets focused on deep learning in search.\n\nThe first dataset was a question answering dataset featuring 100,000 real Bing questions and a human generated answer. \nSince then we released a 1,000,000 question dataset, a natural langauge generation dataset, a passage ranking dataset, \nkeyphrase extraction dataset, crawling dataset, and a conversational search.\n\nThere have been 277 submissions. 20 KeyPhrase Extraction submissions, 87 passage ranking submissions, 0 document ranking \nsubmissions, 73 QnA V2 submissions, 82 NLGEN submisions, and 15 QnA V1 submissions\n\nThis data comes in three tasks/forms: Original QnA dataset(v1.1), Question Answering(v2.1), Natural Language Generation(v2.1). \n\nThe original question answering datset featured 100,000 examples and was released in 2016. Leaderboard is now closed but data is availible below.\n\nThe current competitive tasks are Question Answering and Natural Language Generation. Question Answering features over 1,000,000 queries and \nis much like the original QnA dataset but bigger and with higher quality. The Natural Language Generation dataset features 180,000 examples and \nbuilds upon the QnA dataset to deliver answers that could be spoken by a smart speaker.\n\n\nversion v2.1",
88
+ "citation": "\n@article{DBLP:journals/corr/NguyenRSGTMD16,\n author = {Tri Nguyen and\n Mir Rosenberg and\n Xia Song and\n Jianfeng Gao and\n Saurabh Tiwary and\n Rangan Majumder and\n Li Deng},\n title = {{MS} {MARCO:} {A} Human Generated MAchine Reading COmprehension Dataset},\n journal = {CoRR},\n volume = {abs/1611.09268},\n year = {2016},\n url = {http://arxiv.org/abs/1611.09268},\n archivePrefix = {arXiv},\n eprint = {1611.09268},\n timestamp = {Mon, 13 Aug 2018 16:49:03 +0200},\n biburl = {https://dblp.org/rec/journals/corr/NguyenRSGTMD16.bib},\n bibsource = {dblp computer science bibliography, https://dblp.org}\n}\n}\n",
89
+ "homepage": "https://microsoft.github.io/msmarco/",
90
+ "license": "",
91
+ "features": {
92
+ "answers": {
93
+ "feature": {
94
+ "dtype": "string",
95
+ "id": null,
96
+ "_type": "Value"
97
+ },
98
+ "length": -1,
99
+ "id": null,
100
+ "_type": "Sequence"
101
+ },
102
+ "passages": {
103
+ "feature": {
104
+ "is_selected": {
105
+ "dtype": "int32",
106
+ "id": null,
107
+ "_type": "Value"
108
+ },
109
+ "passage_text": {
110
+ "dtype": "string",
111
+ "id": null,
112
+ "_type": "Value"
113
+ },
114
+ "url": {
115
+ "dtype": "string",
116
+ "id": null,
117
+ "_type": "Value"
118
+ }
119
+ },
120
+ "length": -1,
121
+ "id": null,
122
+ "_type": "Sequence"
123
+ },
124
+ "query": {
125
+ "dtype": "string",
126
+ "id": null,
127
+ "_type": "Value"
128
+ },
129
+ "query_id": {
130
+ "dtype": "int32",
131
+ "id": null,
132
+ "_type": "Value"
133
+ },
134
+ "query_type": {
135
+ "dtype": "string",
136
+ "id": null,
137
+ "_type": "Value"
138
+ },
139
+ "wellFormedAnswers": {
140
+ "feature": {
141
+ "dtype": "string",
142
+ "id": null,
143
+ "_type": "Value"
144
+ },
145
+ "length": -1,
146
+ "id": null,
147
+ "_type": "Sequence"
148
+ }
149
+ },
150
+ "supervised_keys": null,
151
+ "builder_name": "ms_marco",
152
+ "config_name": "v2.1",
153
+ "version": {
154
+ "version_str": "2.1.0",
155
+ "description": "",
156
+ "datasets_version_to_prepare": null,
157
+ "major": 2,
158
+ "minor": 1,
159
+ "patch": 0
160
+ },
161
+ "splits": {
162
+ "validation": {
163
+ "name": "validation",
164
+ "num_bytes": 414286005,
165
+ "num_examples": 101093,
166
+ "dataset_name": "ms_marco"
167
+ },
168
+ "train": {
169
+ "name": "train",
170
+ "num_bytes": 3466972085,
171
+ "num_examples": 808731,
172
+ "dataset_name": "ms_marco"
173
+ },
174
+ "test": {
175
+ "name": "test",
176
+ "num_bytes": 406197152,
177
+ "num_examples": 101092,
178
+ "dataset_name": "ms_marco"
179
+ }
180
+ },
181
+ "download_checksums": {
182
+ "https://msmarco.blob.core.windows.net/msmarco/train_v2.1.json.gz": {
183
+ "num_bytes": 1112116929,
184
+ "checksum": "e91745411ca81e441a3bb75deb71ce000dc2fc31334085b7d499982f14218fe2"
185
+ },
186
+ "https://msmarco.blob.core.windows.net/msmarco/dev_v2.1.json.gz": {
187
+ "num_bytes": 138303699,
188
+ "checksum": "5b3c9c20d1808ee199a930941b0d96f79e397e9234f77a1496890b138df7cb3c"
189
+ },
190
+ "https://msmarco.blob.core.windows.net/msmarco/eval_v2.1_public.json.gz": {
191
+ "num_bytes": 133851237,
192
+ "checksum": "05ac0e448450d507e7ff8e37f48a41cc2d015f5bd2c7974d2445f00a53625db6"
193
+ }
194
+ },
195
+ "download_size": 1384271865,
196
+ "dataset_size": 4287455242,
197
+ "size_in_bytes": 5671727107
198
+ }
199
+ }
v1.1/test-00000-of-00001.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:28721adaa7767035faa26c647052db537ecc0675ccd85c6d3a559a157bb14f82
3
+ size 20484569
v1.1/train-00000-of-00001.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:11d5e51d70ce87b47b1c132ef7c55f4cfb93b22c122058aa0591a65c4c0dc1f1
3
+ size 175452226
v1.1/validation-00000-of-00001.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:67aca2e1b2fffe418a6bd317142887c90973c9c98dd8d940eaae3f8521daed7c
3
+ size 21391358