bboldt commited on
Commit
0e8a9b5
·
1 Parent(s): a6b99c2

Add corpus metadata to Croissant

Browse files
Files changed (2) hide show
  1. croissant.json +263 -7
  2. util/croissant.py +99 -4
croissant.json CHANGED
@@ -49,7 +49,7 @@
49
  "description": "ELCC is a collection of emergent language corpora with accompanying metadata and analyses.",
50
  "conformsTo": "http://mlcommons.org/croissant/1.0",
51
  "citeAs": "@misc{boldt2024elcc,\n title = \"{ELCC}: the Emergent Language Corpus Collection\",\n author = \"Brendon Boldt and David Mortensen\",\n year = 2024,\n url = \"https://huggingface.co/datasets/bboldt/elcc\",\n}",
52
- "datePublished": "2024-06-13 06:42:12.302053+00:00",
53
  "keywords": [
54
  "emergent communication",
55
  "emergent language",
@@ -65,7 +65,7 @@
65
  "name": "repo",
66
  "contentUrl": "https://huggingface.co/datasets/bboldt/elcc/",
67
  "encodingFormat": "git+https",
68
- "sha256": "b3c660db198ddca4cf1de63a4dbb0ecb573d28a3"
69
  },
70
  {
71
  "@type": "cr:FileSet",
@@ -87,6 +87,26 @@
87
  "encodingFormat": "text/plain",
88
  "includes": "systems/*/system.json"
89
  },
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
90
  {
91
  "@type": "cr:FileObject",
92
  "@id": "systems/babyai-sr/data/GoToObj/corpus.json",
@@ -829,11 +849,6 @@
829
  "@id": "system-metadata-raw/path",
830
  "name": "path",
831
  "dataType": "sc:Text",
832
- "references": {
833
- "field": {
834
- "@id": "system-metadata/path"
835
- }
836
- },
837
  "source": {
838
  "fileSet": {
839
  "@id": "system-metadata-files-raw"
@@ -1076,6 +1091,247 @@
1076
  }
1077
  ]
1078
  },
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1079
  {
1080
  "@type": "cr:RecordSet",
1081
  "@id": "babyai-sr/GoToObj",
 
49
  "description": "ELCC is a collection of emergent language corpora with accompanying metadata and analyses.",
50
  "conformsTo": "http://mlcommons.org/croissant/1.0",
51
  "citeAs": "@misc{boldt2024elcc,\n title = \"{ELCC}: the Emergent Language Corpus Collection\",\n author = \"Brendon Boldt and David Mortensen\",\n year = 2024,\n url = \"https://huggingface.co/datasets/bboldt/elcc\",\n}",
52
+ "datePublished": "2024-06-13 07:19:21.844134+00:00",
53
  "keywords": [
54
  "emergent communication",
55
  "emergent language",
 
65
  "name": "repo",
66
  "contentUrl": "https://huggingface.co/datasets/bboldt/elcc/",
67
  "encodingFormat": "git+https",
68
+ "sha256": "main"
69
  },
70
  {
71
  "@type": "cr:FileSet",
 
87
  "encodingFormat": "text/plain",
88
  "includes": "systems/*/system.json"
89
  },
90
+ {
91
+ "@type": "cr:FileSet",
92
+ "@id": "corpus-metadata-files",
93
+ "name": "corpus-metadata-files",
94
+ "containedIn": {
95
+ "@id": "repo"
96
+ },
97
+ "encodingFormat": "application/json",
98
+ "includes": "systems/*/data/*/metadata.json"
99
+ },
100
+ {
101
+ "@type": "cr:FileSet",
102
+ "@id": "corpus-metadata-files-raw",
103
+ "name": "corpus-metadata-files-raw",
104
+ "containedIn": {
105
+ "@id": "repo"
106
+ },
107
+ "encodingFormat": "text/plain",
108
+ "includes": "systems/*/data/*/metadata.json"
109
+ },
110
  {
111
  "@type": "cr:FileObject",
112
  "@id": "systems/babyai-sr/data/GoToObj/corpus.json",
 
849
  "@id": "system-metadata-raw/path",
850
  "name": "path",
851
  "dataType": "sc:Text",
 
 
 
 
 
852
  "source": {
853
  "fileSet": {
854
  "@id": "system-metadata-files-raw"
 
1091
  }
1092
  ]
1093
  },
1094
+ {
1095
+ "@type": "cr:RecordSet",
1096
+ "@id": "corpus-metadata-raw",
1097
+ "name": "corpus-metadata-raw",
1098
+ "field": [
1099
+ {
1100
+ "@type": "cr:Field",
1101
+ "@id": "corpus-metadata-raw/path",
1102
+ "name": "path",
1103
+ "dataType": "sc:Text",
1104
+ "source": {
1105
+ "fileSet": {
1106
+ "@id": "corpus-metadata-files-raw"
1107
+ },
1108
+ "extract": {
1109
+ "fileProperty": "fullpath"
1110
+ }
1111
+ }
1112
+ },
1113
+ {
1114
+ "@type": "cr:Field",
1115
+ "@id": "corpus-metadata-raw/raw",
1116
+ "name": "raw",
1117
+ "dataType": "sc:Text",
1118
+ "source": {
1119
+ "fileSet": {
1120
+ "@id": "corpus-metadata-files-raw"
1121
+ },
1122
+ "extract": {
1123
+ "fileProperty": "content"
1124
+ }
1125
+ }
1126
+ }
1127
+ ]
1128
+ },
1129
+ {
1130
+ "@type": "cr:RecordSet",
1131
+ "@id": "corpus-metadata",
1132
+ "name": "corpus-metadata",
1133
+ "field": [
1134
+ {
1135
+ "@type": "cr:Field",
1136
+ "@id": "corpus-metadata/path",
1137
+ "name": "path",
1138
+ "dataType": "sc:Text",
1139
+ "source": {
1140
+ "fileSet": {
1141
+ "@id": "corpus-metadata-files"
1142
+ },
1143
+ "extract": {
1144
+ "fileProperty": "fullpath"
1145
+ },
1146
+ "transform": {
1147
+ "replace": "a"
1148
+ }
1149
+ }
1150
+ },
1151
+ {
1152
+ "@type": "cr:Field",
1153
+ "@id": "corpus-metadata/metrics.analysis.token_count",
1154
+ "name": "metrics.analysis.token_count",
1155
+ "dataType": "sc:Integer",
1156
+ "source": {
1157
+ "fileSet": {
1158
+ "@id": "corpus-metadata-files"
1159
+ },
1160
+ "extract": {
1161
+ "jsonPath": "$.metrics.analysis[\"Token Count\"]"
1162
+ }
1163
+ }
1164
+ },
1165
+ {
1166
+ "@type": "cr:Field",
1167
+ "@id": "corpus-metadata/metrics.analysis.line_count",
1168
+ "name": "metrics.analysis.line_count",
1169
+ "dataType": "sc:Integer",
1170
+ "source": {
1171
+ "fileSet": {
1172
+ "@id": "corpus-metadata-files"
1173
+ },
1174
+ "extract": {
1175
+ "jsonPath": "$.metrics.analysis[\"Line Count\"]"
1176
+ }
1177
+ }
1178
+ },
1179
+ {
1180
+ "@type": "cr:Field",
1181
+ "@id": "corpus-metadata/metrics.analysis.tokens_per_line",
1182
+ "name": "metrics.analysis.tokens_per_line",
1183
+ "dataType": "sc:Float",
1184
+ "source": {
1185
+ "fileSet": {
1186
+ "@id": "corpus-metadata-files"
1187
+ },
1188
+ "extract": {
1189
+ "jsonPath": "$.metrics.analysis[\"Tokens per Line\"]"
1190
+ }
1191
+ }
1192
+ },
1193
+ {
1194
+ "@type": "cr:Field",
1195
+ "@id": "corpus-metadata/metrics.analysis.tokens_per_line_sd",
1196
+ "name": "metrics.analysis.tokens_per_line_sd",
1197
+ "dataType": "sc:Float",
1198
+ "source": {
1199
+ "fileSet": {
1200
+ "@id": "corpus-metadata-files"
1201
+ },
1202
+ "extract": {
1203
+ "jsonPath": "$.metrics.analysis[\"Tokens per Line SD\"]"
1204
+ }
1205
+ }
1206
+ },
1207
+ {
1208
+ "@type": "cr:Field",
1209
+ "@id": "corpus-metadata/metrics.analysis.unique_tokens",
1210
+ "name": "metrics.analysis.unique_tokens",
1211
+ "dataType": "sc:Integer",
1212
+ "source": {
1213
+ "fileSet": {
1214
+ "@id": "corpus-metadata-files"
1215
+ },
1216
+ "extract": {
1217
+ "jsonPath": "$.metrics.analysis[\"Unique Tokens\"]"
1218
+ }
1219
+ }
1220
+ },
1221
+ {
1222
+ "@type": "cr:Field",
1223
+ "@id": "corpus-metadata/metrics.analysis.unique_lines",
1224
+ "name": "metrics.analysis.unique_lines",
1225
+ "dataType": "sc:Integer",
1226
+ "source": {
1227
+ "fileSet": {
1228
+ "@id": "corpus-metadata-files"
1229
+ },
1230
+ "extract": {
1231
+ "jsonPath": "$.metrics.analysis[\"Unique Lines\"]"
1232
+ }
1233
+ }
1234
+ },
1235
+ {
1236
+ "@type": "cr:Field",
1237
+ "@id": "corpus-metadata/metrics.analysis.1-gram_entropy",
1238
+ "name": "metrics.analysis.1-gram_entropy",
1239
+ "dataType": "sc:Float",
1240
+ "source": {
1241
+ "fileSet": {
1242
+ "@id": "corpus-metadata-files"
1243
+ },
1244
+ "extract": {
1245
+ "jsonPath": "$.metrics.analysis[\"1-gram Entropy\"]"
1246
+ }
1247
+ }
1248
+ },
1249
+ {
1250
+ "@type": "cr:Field",
1251
+ "@id": "corpus-metadata/metrics.analysis.1-gram_normalized_entropy",
1252
+ "name": "metrics.analysis.1-gram_normalized_entropy",
1253
+ "dataType": "sc:Float",
1254
+ "source": {
1255
+ "fileSet": {
1256
+ "@id": "corpus-metadata-files"
1257
+ },
1258
+ "extract": {
1259
+ "jsonPath": "$.metrics.analysis[\"1-gram Normalized Entropy\"]"
1260
+ }
1261
+ }
1262
+ },
1263
+ {
1264
+ "@type": "cr:Field",
1265
+ "@id": "corpus-metadata/metrics.analysis.2-gram_entropy",
1266
+ "name": "metrics.analysis.2-gram_entropy",
1267
+ "dataType": "sc:Float",
1268
+ "source": {
1269
+ "fileSet": {
1270
+ "@id": "corpus-metadata-files"
1271
+ },
1272
+ "extract": {
1273
+ "jsonPath": "$.metrics.analysis[\"2-gram Entropy\"]"
1274
+ }
1275
+ }
1276
+ },
1277
+ {
1278
+ "@type": "cr:Field",
1279
+ "@id": "corpus-metadata/metrics.analysis.2-gram_conditional_entropy",
1280
+ "name": "metrics.analysis.2-gram_conditional_entropy",
1281
+ "dataType": "sc:Float",
1282
+ "source": {
1283
+ "fileSet": {
1284
+ "@id": "corpus-metadata-files"
1285
+ },
1286
+ "extract": {
1287
+ "jsonPath": "$.metrics.analysis[\"2-gram Conditional Entropy\"]"
1288
+ }
1289
+ }
1290
+ },
1291
+ {
1292
+ "@type": "cr:Field",
1293
+ "@id": "corpus-metadata/metrics.analysis.entropy_per_line",
1294
+ "name": "metrics.analysis.entropy_per_line",
1295
+ "dataType": "sc:Float",
1296
+ "source": {
1297
+ "fileSet": {
1298
+ "@id": "corpus-metadata-files"
1299
+ },
1300
+ "extract": {
1301
+ "jsonPath": "$.metrics.analysis[\"Entropy per Line\"]"
1302
+ }
1303
+ }
1304
+ },
1305
+ {
1306
+ "@type": "cr:Field",
1307
+ "@id": "corpus-metadata/metrics.analysis.eos_token_present",
1308
+ "name": "metrics.analysis.eos_token_present",
1309
+ "dataType": "sc:Integer",
1310
+ "source": {
1311
+ "fileSet": {
1312
+ "@id": "corpus-metadata-files"
1313
+ },
1314
+ "extract": {
1315
+ "jsonPath": "$.metrics.analysis[\"EoS Token Present\"]"
1316
+ }
1317
+ }
1318
+ },
1319
+ {
1320
+ "@type": "cr:Field",
1321
+ "@id": "corpus-metadata/metrics.analysis.eos_padding",
1322
+ "name": "metrics.analysis.eos_padding",
1323
+ "dataType": "sc:Integer",
1324
+ "source": {
1325
+ "fileSet": {
1326
+ "@id": "corpus-metadata-files"
1327
+ },
1328
+ "extract": {
1329
+ "jsonPath": "$.metrics.analysis[\"EoS Padding\"]"
1330
+ }
1331
+ }
1332
+ }
1333
+ ]
1334
+ },
1335
  {
1336
  "@type": "cr:RecordSet",
1337
  "@id": "babyai-sr/GoToObj",
util/croissant.py CHANGED
@@ -17,8 +17,7 @@ _distribution: list[mlc.FileObject | mlc.FileSet] = [
17
  name="repo",
18
  content_url="https://huggingface.co/datasets/bboldt/elcc/",
19
  encoding_format="git+https",
20
- # sha256="main",
21
- sha256="b3c660db198ddca4cf1de63a4dbb0ecb573d28a3",
22
  ),
23
  mlc.FileSet(
24
  id="system-metadata-files",
@@ -34,6 +33,20 @@ _distribution: list[mlc.FileObject | mlc.FileSet] = [
34
  includes=["systems/*/system.json"],
35
  encoding_format="text/plain",
36
  ),
 
 
 
 
 
 
 
 
 
 
 
 
 
 
37
  ]
38
 
39
  _metadata = mlc.Metadata(
@@ -196,9 +209,89 @@ def insert_system_md(metadata: mlc.Metadata) -> None:
196
  )
197
 
198
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
199
  @functools.cache
200
  def get_metadata() -> mlc.Metadata:
201
  insert_system_md(_metadata)
 
202
  insert_corpora(_metadata)
203
  return _metadata
204
 
@@ -215,12 +308,14 @@ def save_metadata() -> None:
215
  def test() -> None:
216
  dataset = mlc.Dataset(jsonld="croissant.json")
217
  # records = dataset.records(record_set="babyai-sr/GoToObj")
218
- records = dataset.records(record_set="system-metadata")
 
 
219
 
220
  for i, x in enumerate(records):
221
  print(i, x)
222
  if i > 10:
223
- # break
224
  pass
225
 
226
 
 
17
  name="repo",
18
  content_url="https://huggingface.co/datasets/bboldt/elcc/",
19
  encoding_format="git+https",
20
+ sha256="main",
 
21
  ),
22
  mlc.FileSet(
23
  id="system-metadata-files",
 
33
  includes=["systems/*/system.json"],
34
  encoding_format="text/plain",
35
  ),
36
+ mlc.FileSet(
37
+ id="corpus-metadata-files",
38
+ name="corpus-metadata-files",
39
+ contained_in=["repo"],
40
+ includes=["systems/*/data/*/metadata.json"],
41
+ encoding_format="application/json",
42
+ ),
43
+ mlc.FileSet(
44
+ id="corpus-metadata-files-raw",
45
+ name="corpus-metadata-files-raw",
46
+ contained_in=["repo"],
47
+ includes=["systems/*/data/*/metadata.json"],
48
+ encoding_format="text/plain",
49
+ ),
50
  ]
51
 
52
  _metadata = mlc.Metadata(
 
209
  )
210
 
211
 
212
+ def insert_corpus_md(metadata: mlc.Metadata) -> None:
213
+ metadata.record_sets.append(
214
+ mlc.RecordSet(
215
+ id="corpus-metadata-raw",
216
+ name="corpus-metadata-raw",
217
+ fields=[
218
+ mlc.Field(
219
+ id="corpus-metadata-raw/path",
220
+ name="path",
221
+ data_types=[mlc.DataType.TEXT],
222
+ source=mlc.Source(
223
+ file_set="corpus-metadata-files-raw",
224
+ extract=mlc.Extract(file_property=mlc.FileProperty.fullpath),
225
+ ),
226
+ # References always seem to cause conflicting read method errors.
227
+ # references={"field": {"@id": "corpus-metadata/path"}},
228
+ ),
229
+ mlc.Field(
230
+ id="corpus-metadata-raw/raw",
231
+ name="raw",
232
+ data_types=[mlc.DataType.TEXT],
233
+ source=mlc.Source(
234
+ file_set="corpus-metadata-files-raw",
235
+ extract=mlc.Extract(file_property=mlc.FileProperty.content),
236
+ ),
237
+ ),
238
+ ],
239
+ )
240
+ )
241
+
242
+ exemplar_path = "systems/nav-to-center/data/temperature_10/metadata.json"
243
+ with open(exemplar_path) as fo:
244
+ exemplar_data = json.load(fo)
245
+ fields = [
246
+ mlc.Field(
247
+ id="corpus-metadata/path",
248
+ name="path",
249
+ data_types=[mlc.DataType.TEXT],
250
+ source=mlc.Source(
251
+ file_set="corpus-metadata-files",
252
+ extract=mlc.Extract(file_property=mlc.FileProperty.fullpath),
253
+ # transforms=[mlc.Transform(regex=r"(....)")],
254
+ ),
255
+ # References always seem to cause conflicting read method errors.
256
+ # references={"field": {"@id": "corpus-metadata/path"}},
257
+ ),
258
+ ]
259
+ items = exemplar_data["metrics"]["analysis"].items()
260
+ for k, v in items:
261
+ name = f"metrics.analysis.{k}".replace(" ", "_").lower()
262
+ if isinstance(v, int):
263
+ typ = mlc.DataType.INTEGER
264
+ elif isinstance(v, bool):
265
+ typ = mlc.DataType.BOOL
266
+ else:
267
+ typ = mlc.DataType.FLOAT
268
+ jp = f'$.metrics.analysis["{k}"]'
269
+ fields.append(
270
+ mlc.Field(
271
+ id=f"corpus-metadata/{name}",
272
+ name=name,
273
+ data_types=[typ],
274
+ source=mlc.Source(
275
+ file_set="corpus-metadata-files",
276
+ # extract=mlc.Extract(file_property=mlc.FileProperty.fullpath),
277
+ extract=mlc.Extract(json_path=jp),
278
+ ),
279
+ )
280
+ )
281
+
282
+ metadata.record_sets.append(
283
+ mlc.RecordSet(
284
+ id="corpus-metadata",
285
+ name="corpus-metadata",
286
+ fields=fields,
287
+ )
288
+ )
289
+
290
+
291
  @functools.cache
292
  def get_metadata() -> mlc.Metadata:
293
  insert_system_md(_metadata)
294
+ insert_corpus_md(_metadata)
295
  insert_corpora(_metadata)
296
  return _metadata
297
 
 
308
  def test() -> None:
309
  dataset = mlc.Dataset(jsonld="croissant.json")
310
  # records = dataset.records(record_set="babyai-sr/GoToObj")
311
+ # records = dataset.records(record_set="system-metadata")
312
+ # records = dataset.records(record_set="system-metadata-raw")
313
+ records = dataset.records(record_set="corpus-metadata")
314
 
315
  for i, x in enumerate(records):
316
  print(i, x)
317
  if i > 10:
318
+ break
319
  pass
320
 
321