Muennighoff commited on
Commit
1114f13
1 Parent(s): d4416e0

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +170 -0
README.md CHANGED
@@ -158,3 +158,173 @@ ON commits_table_base.commit = commit_base
158
  Then export the final dataset from GCP to a bucket as parquet files.
159
  Then copy those parquet files to the hf dataset on an instance.
160
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
158
  Then export the final dataset from GCP to a bucket as parquet files.
159
  Then copy those parquet files to the hf dataset on an instance.
160
 
161
+
162
+
163
+
164
+
165
+
166
+ ### Redoing it to add licenses
167
+
168
+
169
+
170
+
171
+ Samples: 76674567
172
+ 76,674,567
173
+
174
+
175
+ SQL queries executed one-by-one. They're split up as BigQuery was raising `Resources exceeded during query execution` when running all in one.
176
+
177
+ ```sql
178
+ SELECT
179
+ commit, difference, subject, message, unnested_repo_name, license
180
+ FROM (
181
+ SELECT
182
+ repo_name,
183
+ lang.name AS language_name
184
+ FROM
185
+ `bigquery-public-data.github_repos.languages` AS lang_table,
186
+ UNNEST(LANGUAGE) AS lang) lang_table
187
+ JOIN
188
+ `bigquery-public-data.github_repos.licenses` AS license_table
189
+ ON
190
+ license_table.repo_name = lang_table.repo_name
191
+ JOIN (
192
+ SELECT
193
+ *
194
+ FROM
195
+ `bigquery-public-data.github_repos.commits` AS commits_table,
196
+ UNNEST(repo_name) AS unnested_repo_name) commits_table
197
+ ON
198
+ commits_table.unnested_repo_name = lang_table.repo_name
199
+ WHERE
200
+ ((license_table.license LIKE 'mit')
201
+ OR (license_table.license LIKE 'artistic-2.0')
202
+ OR (license_table.license LIKE 'isc')
203
+ OR (license_table.license LIKE 'cc0-1.0')
204
+ OR (license_table.license LIKE 'epl-1.0')
205
+ OR (license_table.license LIKE 'mpl-2.0')
206
+ OR (license_table.license LIKE 'unlicense')
207
+ OR (license_table.license LIKE 'apache-2.0')
208
+ OR (license_table.license LIKE 'bsd-3-clause')
209
+ OR (license_table.license LIKE 'agpl-3.0')
210
+ OR (license_table.license LIKE 'lgpl-2.1')
211
+ OR (license_table.license LIKE 'bsd-2-clause'))
212
+ AND ( (lang_table.language_name LIKE 'Python')
213
+ OR (lang_table.language_name LIKE 'Java')
214
+ OR (lang_table.language_name LIKE 'JavaScript')
215
+ OR (lang_table.language_name LIKE 'HTML')
216
+ OR (lang_table.language_name LIKE 'Common Lisp')
217
+ OR (lang_table.language_name LIKE 'Shell')
218
+ OR (lang_table.language_name LIKE 'R')
219
+ OR (lang_table.language_name LIKE 'Perl%')
220
+ OR (lang_table.language_name LIKE 'SQL')
221
+ OR (lang_table.language_name LIKE 'C')
222
+ OR (lang_table.language_name LIKE 'C#')
223
+ OR (lang_table.language_name LIKE 'C++')
224
+ OR (lang_table.language_name LIKE 'TypeScript')
225
+ OR (lang_table.language_name LIKE 'Go')
226
+ OR (lang_table.language_name LIKE 'Rust')
227
+ OR (lang_table.language_name LIKE 'Swift')
228
+ OR (lang_table.language_name LIKE 'PHP')
229
+ OR (lang_table.language_name LIKE 'Dart')
230
+ OR (lang_table.language_name LIKE 'Kotlin')
231
+ OR (lang_table.language_name LIKE 'Matlab')
232
+ OR (lang_table.language_name LIKE 'MATLAB')
233
+ OR (lang_table.language_name LIKE 'Ruby') )
234
+ AND
235
+ LENGTH(commits_table.message) > 5
236
+ AND
237
+ LENGTH(commits_table.message) < 10000
238
+ AND LOWER(commits_table.message) NOT LIKE 'update readme.md'
239
+ AND LOWER(commits_table.message) NOT LIKE 'initial commit'
240
+ AND LOWER(commits_table.message) NOT LIKE 'update'
241
+ AND LOWER(commits_table.message) NOT LIKE 'mirroring from micro.blog.'
242
+ AND LOWER(commits_table.message) NOT LIKE 'update data.json'
243
+ AND LOWER(commits_table.message) NOT LIKE 'update data.js'
244
+ AND LOWER(commits_table.message) NOT LIKE 'add files via upload'
245
+ AND LOWER(commits_table.message) NOT LIKE 'update readme'
246
+ AND LOWER(commits_table.message) NOT LIKE "can't you see i'm updating the time?"
247
+ AND LOWER(commits_table.message) NOT LIKE 'pi push'
248
+ AND LOWER(commits_table.message) NOT LIKE 'dummy'
249
+ AND LOWER(commits_table.message) NOT LIKE 'update index.html'
250
+ AND LOWER(commits_table.message) NOT LIKE 'first commit'
251
+ AND LOWER(commits_table.message) NOT LIKE 'create readme.md'
252
+ AND LOWER(commits_table.message) NOT LIKE 'heartbeat update'
253
+ AND LOWER(commits_table.message) NOT LIKE 'updated readme'
254
+ AND LOWER(commits_table.message) NOT LIKE 'update log'
255
+ AND LOWER(commits_table.message) NOT LIKE 'test'
256
+ AND LOWER(commits_table.message) NOT LIKE 'no message'
257
+ AND LOWER(commits_table.message) NOT LIKE 'readme'
258
+ AND LOWER(commits_table.message) NOT LIKE 'wip'
259
+ AND LOWER(commits_table.message) NOT LIKE 'updates'
260
+ AND LOWER(commits_table.message) NOT LIKE 'first commit'
261
+ AND LOWER(commits_table.message) NOT LIKE 'commit'
262
+ AND LOWER(commits_table.message) NOT LIKE 'update _config.yaml'
263
+ AND LOWER(commits_table.message) NOT LIKE 'update data.json'
264
+ AND LOWER(commits_table.message) NOT LIKE 'update data.js'
265
+ AND LOWER(commits_table.message) NOT LIKE 'merge%';
266
+ ```
267
+
268
+
269
+
270
+ 3,641,694,786
271
+
272
+
273
+ ```sql
274
+ SELECT commit, subject, message, STRING_AGG(unnested_repo_name), license AS repos
275
+ FROM `huggingface-ml.commits_table_24122022.commits_table_base`
276
+ GROUP BY commit, subject, message, license
277
+ ```
278
+
279
+ ```sql
280
+ SELECT *
281
+ FROM (
282
+ SELECT
283
+ commit,subject,message,repos,difference,license
284
+ FROM
285
+ `huggingface-ml.commits_table_24122022.commits_table_dedup` AS commits_table_dedup
286
+ JOIN (
287
+ SELECT
288
+ commit AS commit_base,difference
289
+ FROM
290
+ `bigquery-public-data.github_repos.commits` AS commits_table_base
291
+ ) commits_table_base
292
+ ON
293
+ commits_table_base.commit_base = commits_table_dedup.commit
294
+ )
295
+ ```
296
+
297
+ ```sql
298
+ SELECT
299
+ commit,subject,message,repos,license,d.old_path as old_file,d.new_path as new_file
300
+ FROM
301
+ `huggingface-ml.commits_table_24122022.commits_table_dedup_difference` AS commits_table,
302
+ UNNEST(difference) AS d
303
+ WHERE (d.old_path = d.new_path) AND (d.old_path IS NOT NULL) AND (d.new_path IS NOT NULL)
304
+ ```
305
+
306
+ ```sql
307
+ SELECT commit,repos,licenses
308
+ FROM (
309
+ (
310
+ SELECT commit AS commit_base
311
+ FROM `huggingface-ml.commits_table_24122022.commits_table_dedup_files`
312
+ GROUP BY commit
313
+ HAVING COUNT(*) = 1
314
+ )
315
+ JOIN (
316
+ SELECT
317
+ commit,subject,message,repos,old_file,new_file
318
+ FROM
319
+ `huggingface-ml.commits_table_24122022.commits_table_dedup_files` AS commits_table_base
320
+ ) commits_table_base
321
+ ON commits_table_base.commit = commit_base
322
+ )
323
+ ```
324
+
325
+
326
+ Then export the final dataset from GCP to a bucket as parquet files.
327
+ Then copy those parquet files to the hf dataset on an instance.
328
+
329
+
330
+