pretrain_args.json
8.5 KB · 425 lines · json Raw
1 {
2 "raw_exp_name": "tabstar_production",
3 "tabular_layers": 6,
4 "e5": "intfloat/e5-small-v2",
5 "tabular_encoder_type": "d1",
6 "numbers_fusion": "attention",
7 "base_lr": 5e-05,
8 "bert_lr_factor": 1.0,
9 "weight_decay": 0.001,
10 "weight": "uniform",
11 "numbers_verbalization": "full",
12 "unfreeze_layers": 6,
13 "light": false,
14 "debug": false,
15 "max_epoch_examples": 2048,
16 "datasets": [
17 41164,
18 48,
19 46684,
20 41142,
21 40753,
22 23381,
23 "sukritchatterjee/used-cars-dataset-cardekho/cars_details_merges.csv",
24 46674,
25 507,
26 "rounakbanik/the-movies-dataset/movies_metadata.csv",
27 541,
28 185,
29 44957,
30 41156,
31 46441,
32 1480,
33 40927,
34 44969,
35 42343,
36 5,
37 1487,
38 1503,
39 558,
40 1464,
41 300,
42 46701,
43 44974,
44 44975,
45 "stefanoleone992/filmtv-movies-dataset/filmtv_movies.csv",
46 46606,
47 1049,
48 46708,
49 44964,
50 458,
51 46584,
52 "ajinkyablaze/football-manager-data/dataset.csv",
53 40693,
54 1459,
55 42729,
56 45567,
57 42734,
58 43748,
59 46659,
60 40975,
61 40900,
62 28,
63 40672,
64 "imuhammad/us-2020-presidential-election-speeches/us_2020_election_speeches.csv",
65 44970,
66 505,
67 1485,
68 41169,
69 40677,
70 42806,
71 1567,
72 57,
73 1099,
74 525,
75 46678,
76 45577,
77 1483,
78 25,
79 24,
80 46,
81 "noorrizki/top-korean-drama-list-1500/kdrama_list.csv",
82 2,
83 26,
84 1466,
85 40996,
86 "rtatman/chocolate-bar-ratings/flavors_of_cacao.csv",
87 566,
88 "limtis/wikiliq-dataset/spirits_data.csv",
89 560,
90 43743,
91 45064,
92 43822,
93 44973,
94 563,
95 46658,
96 41161,
97 40985,
98 46719,
99 44146,
100 "http://pages.cs.wisc.edu/~anhai/data/784_data/movies1/csv_files/rotten_tomatoes.csv",
101 940,
102 42731,
103 1046,
104 43660,
105 41159,
106 42570,
107 42563,
108 1510,
109 11,
110 43712,
111 40701,
112 307,
113 43420,
114 45012,
115 554,
116 43595,
117 44231,
118 43435,
119 43344,
120 41162,
121 1457,
122 42712,
123 40983,
124 1111,
125 46359,
126 41990,
127 41143,
128 43643,
129 43766,
130 "ruthgn/beer-profile-and-ratings-data-set/beer_profile_and_ratings.csv",
131 46656,
132 1486,
133 46654,
134 42225,
135 41167,
136 "mattop/nba-draft-basketball-player-data-19892021/nbaplayersdraft.csv",
137 44976,
138 40984,
139 45062,
140 42345,
141 "skamlo/wine-price-on-polish-market/wine.csv",
142 215,
143 46607,
144 41147,
145 44125,
146 546,
147 40922,
148 46669,
149 223,
150 "http://pages.cs.wisc.edu/~anhai/data/784_data/books2/csv_files/goodreads.csv",
151 "mustafaimam/used-car-prices-in-pakistan-2021/Used_car_prices_in_Pakistan_cleaned.csv",
152 4549,
153 4538,
154 516,
155 15,
156 43071,
157 45039,
158 46762,
159 "joshuakalobbowles/vivino-wine-data/vivino.csv",
160 43397,
161 4134,
162 43502,
163 422,
164 46430,
165 3050,
166 41150,
167 46444,
168 1116,
169 44963,
170 4541,
171 201,
172 503,
173 41157,
174 41168,
175 42732,
176 43687,
177 377,
178 42165,
179 "sobhanmoosavi/us-accidents/US_Accidents_March23.csv",
180 1515,
181 666,
182 43342,
183 44156,
184 1169,
185 41165,
186 44227,
187 46646,
188 470,
189 41021,
190 504,
191 574,
192 42895,
193 45560,
194 13,
195 46610,
196 "http://pages.cs.wisc.edu/~anhai/data/784_data/bikes/csv_files/bikewale.csv",
197 1493,
198 1475,
199 451,
200 43588,
201 42194,
202 44962,
203 4534,
204 42746,
205 44994,
206 529,
207 32,
208 40966,
209 40971,
210 181,
211 1494,
212 43838,
213 43389,
214 46660,
215 42125,
216 334,
217 1067,
218 188,
219 46603,
220 40705,
221 45547,
222 45038,
223 40945,
224 1495,
225 7,
226 43635,
227 41158,
228 46709,
229 41163,
230 6,
231 564,
232 416,
233 46467,
234 45950,
235 46667,
236 41138,
237 43157,
238 372,
239 "ngshiheng/michelin-guide-restaurants-2021/michelin_my_maps.csv",
240 43160,
241 43454,
242 "https://opendata.vancouver.ca/api/records/1.0/download/?dataset=employee-remuneration-and-expenses-earning-over-75000&format=csv",
243 46726,
244 40978,
245 42572,
246 46372,
247 1461,
248 44228,
249 42571,
250 46664,
251 46328,
252 54,
253 40498,
254 42730,
255 43337,
256 1489,
257 "markusschmitz/museums/museums_prep.csv",
258 1507,
259 43544,
260 46681,
261 469,
262 531,
263 "himanshupoddar/zomato-bangalore-restaurants/zomato.csv",
264 42727,
265 46676,
266 42705,
267 42733,
268 40650,
269 44960,
270 46686,
271 44157,
272 43044,
273 44983,
274 "https://ai-jobs.net/salaries/download/salaries.csv",
275 44984,
276 "peopledatalabssf/free-7-million-company-dataset/companies_sorted.csv",
277 1596,
278 "omkarsabnis/yelp-reviews-dataset/yelp.csv",
279 40981,
280 56,
281 344,
282 44978,
283 1039,
284 311,
285 "verracodeguacas/clear-corpus/CLEAR.csv",
286 46661,
287 46604,
288 549,
289 4552,
290 1128,
291 44959,
292 1547,
293 40685,
294 43826,
295 41540,
296 14,
297 43622,
298 "gregorut/videogamesales/vgsales.csv",
299 46369,
300 41210,
301 42688,
302 46605,
303 40670,
304 46280,
305 40994,
306 951,
307 46640,
308 43479,
309 44993,
310 44230,
311 43573,
312 455,
313 46653,
314 41991,
315 46683,
316 41144,
317 40536,
318 44981,
319 41146,
320 46655,
321 46662,
322 22,
323 46721,
324 31,
325 46663,
326 41972,
327 43672,
328 216,
329 43721,
330 44965,
331 46648,
332 310,
333 1471,
334 "http://pages.cs.wisc.edu/~anhai/data/784_data/baby_products/csv_files/babies_r_us.csv",
335 23,
336 46665,
337 46592,
338 23517,
339 "eilamshapira/human-choice-prediction-in-language-based-games/OPE_train.csv",
340 46598,
341 279,
342 44145,
343 42164,
344 44,
345 43648,
346 40668,
347 45545,
348 547,
349 43618,
350 "neilcosgrove/scotch-whiskey-reviews-update-2020/scotch_review2020.csv",
351 10,
352 "maharshipandya/-spotify-tracks-dataset/dataset.csv",
353 44958,
354 30,
355 46254,
356 1479,
357 "albenft/game-of-thrones-script-all-seasons/Game_of_Thrones_Script.csv",
358 42726,
359 46585,
360 45554,
361 44226,
362 44980,
363 679,
364 "bogdansorin/second-hand-mercedes-benz-registered-2000-2023-ita/mercedes-benz.csv",
365 43463,
366 45955,
367 50,
368 375,
369 23380,
370 1497,
371 1509,
372 "turkibintalib/saudi-arabia-used-cars-dataset/UsedCarsSA_Clean_EN.csv",
373 41980,
374 1542,
375 "ankanhore545/top-ramen-ratings-2022/Top Ramen Ratings .csv",
376 44992,
377 40982,
378 44966,
379 1590,
380 43551,
381 550,
382 42769,
383 44977,
384 43584,
385 3,
386 41166,
387 43072,
388 1511,
389 4135,
390 46668,
391 "hanifalirsyad/coffee-scrap-coffeereview/coffee_clean.csv",
392 46677,
393 44223,
394 "https://www.scimagojr.com/journalrank.php?out=xls",
395 46597,
396 1502,
397 43395,
398 41027,
399 46692,
400 1468,
401 43466,
402 42,
403 60,
404 "hernan4444/animeplanet-recommendation-database-2020/anime.csv",
405 42972,
406 46652,
407 41145,
408 3277,
409 43893,
410 6332,
411 46651,
412 44967,
413 42931,
414 42742,
415 12,
416 43653
417 ],
418 "scheduler": "cycle",
419 "fold": null,
420 "num_folds": 5,
421 "p_mask": 0,
422 "full_exp_name": "2025_05_16__tabstar_production__data_400__tab_6__e5_small__axial_d1__fusion_attention__layers_6__num_verb_full__scheduler_cycle__lr_5e-05__bert_lr_10__wd_0001__mask_0__weight_uniform__git_51b5fd0",
423 "cached": false,
424 "num_datasets": 400
425 }