Hugging Face's logo

langtech-innovation
/
mRoBERTA_retrieval

Sentence Similarity
sentence-transformers
Safetensors
Catalan
Spanish
English
roberta
feature-extraction
Generated from Trainer
dataset_size:65360
loss:ContrastiveLoss
Eval Results (legacy)
text-embeddings-inference
Model card Files
xet
 Community
crodri commited on
Commit
98b312b
·
verified ·
1 Parent(s): 3b1f111

Upload 16 files

Browse files
Files changed (9) hide show
  1. config.json +1 -1
  2. model.safetensors +1 -1
  3. optimizer.pt +3 -0
  4. rng_state.pth +3 -0
  5. scaler.pt +3 -0
  6. scheduler.pt +3 -0
  7. tokenizer.json +2 -2
  8. trainer_state.json +470 -0
  9. training_args.bin +3 -0
config.json CHANGED
@@ -1,5 +1,5 @@
1
  {
2
- "_name_or_path": "output/training_trilingual_query_relevance_BSC-LT-mRoBERTa-2025-03-20_21-52-43/checkpoint-6200/",
3
  "architectures": [
4
  "RobertaModel"
5
  ],
 
1
  {
2
+ "_name_or_path": "BSC-LT/mRoBERTa",
3
  "architectures": [
4
  "RobertaModel"
5
  ],
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:61ece44ba7f8ee74d751061dc315930e9716ac0a0defd0d3fa2ba6c50b118681
3
  size 1130622968
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1c31484f587e2dfb02a6788f0041913015b9763e022321d793af1f1cde903603
3
  size 1130622968
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c6487d4ea1e666b0cc7f296576292298cbba1db1f8d2e72dfadd614b9814b4c0
3
+ size 2256641018
rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:76ff761761ea8118d9aa5b55aea4c1d175f3994942db524819cb517bf8c54004
3
+ size 14244
scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7ad467126d85159780691480b4a33aa1ef0e4ca019776ab21241c5cb6f56ab70
3
+ size 988
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:32420dd7c09d65fb2dc765b58b29995b23996ac6e5524cb982eac495a5dacbd1
3
+ size 1064
tokenizer.json CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:9f2bf5621066c920a83c8b0f75b8d0d95b8b1bad04ac245c6746e6d2b303b76b
3
- size 37007530
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8ed1b089fe336cd80b4fb78f963608799612a0e72d15b0eae7aeb3c7cbb406fc
3
+ size 37007782
trainer_state.json ADDED
@@ -0,0 +1,470 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.009580136276781559,
3
+ "best_model_checkpoint": "output/second_training_trilingual_query_relevance_BSC-LT-mRoBERTa-2025-03-22_09-46-32/checkpoint-1900",
4
+ "epoch": 0.06205905408936504,
5
+ "eval_steps": 100,
6
+ "global_step": 1900,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.003266266004703423,
13
+ "grad_norm": 0.3973948657512665,
14
+ "learning_rate": 8.164598301763554e-07,
15
+ "loss": 0.0264,
16
+ "step": 100
17
+ },
18
+ {
19
+ "epoch": 0.003266266004703423,
20
+ "eval_loss": 0.023852620273828506,
21
+ "eval_runtime": 207.3804,
22
+ "eval_samples_per_second": 74.009,
23
+ "eval_steps_per_second": 37.004,
24
+ "eval_trilingual_query_relevance_dev_cosine_accuracy": 0.7612066718790722,
25
+ "eval_trilingual_query_relevance_dev_cosine_accuracy_threshold": 0.7759683728218079,
26
+ "eval_trilingual_query_relevance_dev_cosine_ap": 0.8577521029975277,
27
+ "eval_trilingual_query_relevance_dev_cosine_f1": 0.745243263950964,
28
+ "eval_trilingual_query_relevance_dev_cosine_f1_threshold": 0.7498887181282043,
29
+ "eval_trilingual_query_relevance_dev_cosine_mcc": 0.48046492229033494,
30
+ "eval_trilingual_query_relevance_dev_cosine_precision": 0.7305958938407612,
31
+ "eval_trilingual_query_relevance_dev_cosine_recall": 0.760489966119364,
32
+ "step": 100
33
+ },
34
+ {
35
+ "epoch": 0.006532532009406846,
36
+ "grad_norm": 0.5370780825614929,
37
+ "learning_rate": 1.632919660352711e-06,
38
+ "loss": 0.0256,
39
+ "step": 200
40
+ },
41
+ {
42
+ "epoch": 0.006532532009406846,
43
+ "eval_loss": 0.022389940917491913,
44
+ "eval_runtime": 208.6146,
45
+ "eval_samples_per_second": 73.571,
46
+ "eval_steps_per_second": 36.786,
47
+ "eval_trilingual_query_relevance_dev_cosine_accuracy": 0.7924159499609069,
48
+ "eval_trilingual_query_relevance_dev_cosine_accuracy_threshold": 0.7661597728729248,
49
+ "eval_trilingual_query_relevance_dev_cosine_ap": 0.8853502724002089,
50
+ "eval_trilingual_query_relevance_dev_cosine_f1": 0.7805273833671399,
51
+ "eval_trilingual_query_relevance_dev_cosine_f1_threshold": 0.7519233226776123,
52
+ "eval_trilingual_query_relevance_dev_cosine_mcc": 0.5785447562424102,
53
+ "eval_trilingual_query_relevance_dev_cosine_precision": 0.8111298482293423,
54
+ "eval_trilingual_query_relevance_dev_cosine_recall": 0.7521501172791243,
55
+ "step": 200
56
+ },
57
+ {
58
+ "epoch": 0.00979879801411027,
59
+ "grad_norm": 0.32472163438796997,
60
+ "learning_rate": 2.4493794905290662e-06,
61
+ "loss": 0.0222,
62
+ "step": 300
63
+ },
64
+ {
65
+ "epoch": 0.00979879801411027,
66
+ "eval_loss": 0.01848435029387474,
67
+ "eval_runtime": 209.1593,
68
+ "eval_samples_per_second": 73.379,
69
+ "eval_steps_per_second": 36.69,
70
+ "eval_trilingual_query_relevance_dev_cosine_accuracy": 0.8763356789158196,
71
+ "eval_trilingual_query_relevance_dev_cosine_accuracy_threshold": 0.7009493112564087,
72
+ "eval_trilingual_query_relevance_dev_cosine_ap": 0.947311702044651,
73
+ "eval_trilingual_query_relevance_dev_cosine_f1": 0.872754089568249,
74
+ "eval_trilingual_query_relevance_dev_cosine_f1_threshold": 0.7001018524169922,
75
+ "eval_trilingual_query_relevance_dev_cosine_mcc": 0.7538668122282384,
76
+ "eval_trilingual_query_relevance_dev_cosine_precision": 0.8987848660590997,
77
+ "eval_trilingual_query_relevance_dev_cosine_recall": 0.8481886890800104,
78
+ "step": 300
79
+ },
80
+ {
81
+ "epoch": 0.013065064018813691,
82
+ "grad_norm": 0.7941808700561523,
83
+ "learning_rate": 3.265839320705422e-06,
84
+ "loss": 0.0181,
85
+ "step": 400
86
+ },
87
+ {
88
+ "epoch": 0.013065064018813691,
89
+ "eval_loss": 0.015285510569810867,
90
+ "eval_runtime": 209.3904,
91
+ "eval_samples_per_second": 73.298,
92
+ "eval_steps_per_second": 36.649,
93
+ "eval_trilingual_query_relevance_dev_cosine_accuracy": 0.9108027104508731,
94
+ "eval_trilingual_query_relevance_dev_cosine_accuracy_threshold": 0.670626699924469,
95
+ "eval_trilingual_query_relevance_dev_cosine_ap": 0.9678805741968552,
96
+ "eval_trilingual_query_relevance_dev_cosine_f1": 0.9096063822773126,
97
+ "eval_trilingual_query_relevance_dev_cosine_f1_threshold": 0.6668647527694702,
98
+ "eval_trilingual_query_relevance_dev_cosine_mcc": 0.8215733554132029,
99
+ "eval_trilingual_query_relevance_dev_cosine_precision": 0.9205925530495129,
100
+ "eval_trilingual_query_relevance_dev_cosine_recall": 0.8988793328120928,
101
+ "step": 400
102
+ },
103
+ {
104
+ "epoch": 0.016331330023517117,
105
+ "grad_norm": 0.5087969303131104,
106
+ "learning_rate": 4.082299150881777e-06,
107
+ "loss": 0.0161,
108
+ "step": 500
109
+ },
110
+ {
111
+ "epoch": 0.016331330023517117,
112
+ "eval_loss": 0.01322067528963089,
113
+ "eval_runtime": 209.5742,
114
+ "eval_samples_per_second": 73.234,
115
+ "eval_steps_per_second": 36.617,
116
+ "eval_trilingual_query_relevance_dev_cosine_accuracy": 0.911128485796195,
117
+ "eval_trilingual_query_relevance_dev_cosine_accuracy_threshold": 0.739102303981781,
118
+ "eval_trilingual_query_relevance_dev_cosine_ap": 0.9677564397514283,
119
+ "eval_trilingual_query_relevance_dev_cosine_f1": 0.910314481687589,
120
+ "eval_trilingual_query_relevance_dev_cosine_f1_threshold": 0.7265805006027222,
121
+ "eval_trilingual_query_relevance_dev_cosine_mcc": 0.8194683277198579,
122
+ "eval_trilingual_query_relevance_dev_cosine_precision": 0.9041131105398458,
123
+ "eval_trilingual_query_relevance_dev_cosine_recall": 0.9166015115976023,
124
+ "step": 500
125
+ },
126
+ {
127
+ "epoch": 0.01959759602822054,
128
+ "grad_norm": 0.36360979080200195,
129
+ "learning_rate": 4.8987589810581325e-06,
130
+ "loss": 0.0142,
131
+ "step": 600
132
+ },
133
+ {
134
+ "epoch": 0.01959759602822054,
135
+ "eval_loss": 0.01228660810738802,
136
+ "eval_runtime": 209.2625,
137
+ "eval_samples_per_second": 73.343,
138
+ "eval_steps_per_second": 36.672,
139
+ "eval_trilingual_query_relevance_dev_cosine_accuracy": 0.913408913213448,
140
+ "eval_trilingual_query_relevance_dev_cosine_accuracy_threshold": 0.6841300129890442,
141
+ "eval_trilingual_query_relevance_dev_cosine_ap": 0.9697296920117046,
142
+ "eval_trilingual_query_relevance_dev_cosine_f1": 0.913040633383498,
143
+ "eval_trilingual_query_relevance_dev_cosine_f1_threshold": 0.6841300129890442,
144
+ "eval_trilingual_query_relevance_dev_cosine_mcc": 0.8268474874637588,
145
+ "eval_trilingual_query_relevance_dev_cosine_precision": 0.9169404652385333,
146
+ "eval_trilingual_query_relevance_dev_cosine_recall": 0.9091738337242637,
147
+ "step": 600
148
+ },
149
+ {
150
+ "epoch": 0.02286386203292396,
151
+ "grad_norm": 0.798312246799469,
152
+ "learning_rate": 5.715218811234488e-06,
153
+ "loss": 0.0141,
154
+ "step": 700
155
+ },
156
+ {
157
+ "epoch": 0.02286386203292396,
158
+ "eval_loss": 0.012058778665959835,
159
+ "eval_runtime": 209.6366,
160
+ "eval_samples_per_second": 73.212,
161
+ "eval_steps_per_second": 36.606,
162
+ "eval_trilingual_query_relevance_dev_cosine_accuracy": 0.9113891060724524,
163
+ "eval_trilingual_query_relevance_dev_cosine_accuracy_threshold": 0.6641287803649902,
164
+ "eval_trilingual_query_relevance_dev_cosine_ap": 0.9676510402408733,
165
+ "eval_trilingual_query_relevance_dev_cosine_f1": 0.9126263272355123,
166
+ "eval_trilingual_query_relevance_dev_cosine_f1_threshold": 0.6603518724441528,
167
+ "eval_trilingual_query_relevance_dev_cosine_mcc": 0.8225678050449036,
168
+ "eval_trilingual_query_relevance_dev_cosine_precision": 0.8962311557788945,
169
+ "eval_trilingual_query_relevance_dev_cosine_recall": 0.9296325254104769,
170
+ "step": 700
171
+ },
172
+ {
173
+ "epoch": 0.026130128037627383,
174
+ "grad_norm": 0.46033045649528503,
175
+ "learning_rate": 6.531678641410844e-06,
176
+ "loss": 0.011,
177
+ "step": 800
178
+ },
179
+ {
180
+ "epoch": 0.026130128037627383,
181
+ "eval_loss": 0.011666857637465,
182
+ "eval_runtime": 209.6251,
183
+ "eval_samples_per_second": 73.216,
184
+ "eval_steps_per_second": 36.608,
185
+ "eval_trilingual_query_relevance_dev_cosine_accuracy": 0.9096950742767788,
186
+ "eval_trilingual_query_relevance_dev_cosine_accuracy_threshold": 0.7075139284133911,
187
+ "eval_trilingual_query_relevance_dev_cosine_ap": 0.9668207553672179,
188
+ "eval_trilingual_query_relevance_dev_cosine_f1": 0.9088757396449704,
189
+ "eval_trilingual_query_relevance_dev_cosine_f1_threshold": 0.707144558429718,
190
+ "eval_trilingual_query_relevance_dev_cosine_mcc": 0.8195226683234668,
191
+ "eval_trilingual_query_relevance_dev_cosine_precision": 0.9171974522292994,
192
+ "eval_trilingual_query_relevance_dev_cosine_recall": 0.9007036747458952,
193
+ "step": 800
194
+ },
195
+ {
196
+ "epoch": 0.02939639404233081,
197
+ "grad_norm": 0.3261055648326874,
198
+ "learning_rate": 7.348138471587198e-06,
199
+ "loss": 0.0107,
200
+ "step": 900
201
+ },
202
+ {
203
+ "epoch": 0.02939639404233081,
204
+ "eval_loss": 0.011733059771358967,
205
+ "eval_runtime": 209.5431,
206
+ "eval_samples_per_second": 73.245,
207
+ "eval_steps_per_second": 36.623,
208
+ "eval_trilingual_query_relevance_dev_cosine_accuracy": 0.9001172791243158,
209
+ "eval_trilingual_query_relevance_dev_cosine_accuracy_threshold": 0.762702226638794,
210
+ "eval_trilingual_query_relevance_dev_cosine_ap": 0.9633731347705192,
211
+ "eval_trilingual_query_relevance_dev_cosine_f1": 0.9008349389852279,
212
+ "eval_trilingual_query_relevance_dev_cosine_f1_threshold": 0.7406441569328308,
213
+ "eval_trilingual_query_relevance_dev_cosine_mcc": 0.7991356063835129,
214
+ "eval_trilingual_query_relevance_dev_cosine_precision": 0.8881712259371833,
215
+ "eval_trilingual_query_relevance_dev_cosine_recall": 0.9138649986968986,
216
+ "step": 900
217
+ },
218
+ {
219
+ "epoch": 0.032662660047034234,
220
+ "grad_norm": 0.20871874690055847,
221
+ "learning_rate": 8.164598301763554e-06,
222
+ "loss": 0.0121,
223
+ "step": 1000
224
+ },
225
+ {
226
+ "epoch": 0.032662660047034234,
227
+ "eval_loss": 0.012324756942689419,
228
+ "eval_runtime": 209.7614,
229
+ "eval_samples_per_second": 73.169,
230
+ "eval_steps_per_second": 36.584,
231
+ "eval_trilingual_query_relevance_dev_cosine_accuracy": 0.9081313526192337,
232
+ "eval_trilingual_query_relevance_dev_cosine_accuracy_threshold": 0.652023434638977,
233
+ "eval_trilingual_query_relevance_dev_cosine_ap": 0.965490592482477,
234
+ "eval_trilingual_query_relevance_dev_cosine_f1": 0.908167252833138,
235
+ "eval_trilingual_query_relevance_dev_cosine_f1_threshold": 0.652023434638977,
236
+ "eval_trilingual_query_relevance_dev_cosine_mcc": 0.8162629547318717,
237
+ "eval_trilingual_query_relevance_dev_cosine_precision": 0.9078125,
238
+ "eval_trilingual_query_relevance_dev_cosine_recall": 0.90852228303362,
239
+ "step": 1000
240
+ },
241
+ {
242
+ "epoch": 0.035928926051737656,
243
+ "grad_norm": 0.6639214158058167,
244
+ "learning_rate": 8.98105813193991e-06,
245
+ "loss": 0.0136,
246
+ "step": 1100
247
+ },
248
+ {
249
+ "epoch": 0.035928926051737656,
250
+ "eval_loss": 0.01079597044736147,
251
+ "eval_runtime": 209.7205,
252
+ "eval_samples_per_second": 73.183,
253
+ "eval_steps_per_second": 36.592,
254
+ "eval_trilingual_query_relevance_dev_cosine_accuracy": 0.9102814698983581,
255
+ "eval_trilingual_query_relevance_dev_cosine_accuracy_threshold": 0.7254592180252075,
256
+ "eval_trilingual_query_relevance_dev_cosine_ap": 0.9699201116888946,
257
+ "eval_trilingual_query_relevance_dev_cosine_f1": 0.9100495081334791,
258
+ "eval_trilingual_query_relevance_dev_cosine_f1_threshold": 0.7066939473152161,
259
+ "eval_trilingual_query_relevance_dev_cosine_mcc": 0.8179880332203355,
260
+ "eval_trilingual_query_relevance_dev_cosine_precision": 0.8982104327960401,
261
+ "eval_trilingual_query_relevance_dev_cosine_recall": 0.9222048475371384,
262
+ "step": 1100
263
+ },
264
+ {
265
+ "epoch": 0.03919519205644108,
266
+ "grad_norm": 0.1763276755809784,
267
+ "learning_rate": 9.797517962116265e-06,
268
+ "loss": 0.0123,
269
+ "step": 1200
270
+ },
271
+ {
272
+ "epoch": 0.03919519205644108,
273
+ "eval_loss": 0.010449341498315334,
274
+ "eval_runtime": 209.7233,
275
+ "eval_samples_per_second": 73.182,
276
+ "eval_steps_per_second": 36.591,
277
+ "eval_trilingual_query_relevance_dev_cosine_accuracy": 0.9190122491529841,
278
+ "eval_trilingual_query_relevance_dev_cosine_accuracy_threshold": 0.6978535652160645,
279
+ "eval_trilingual_query_relevance_dev_cosine_ap": 0.9728747278408176,
280
+ "eval_trilingual_query_relevance_dev_cosine_f1": 0.9192095885973437,
281
+ "eval_trilingual_query_relevance_dev_cosine_f1_threshold": 0.6935862302780151,
282
+ "eval_trilingual_query_relevance_dev_cosine_mcc": 0.837557083900232,
283
+ "eval_trilingual_query_relevance_dev_cosine_precision": 0.914057466821286,
284
+ "eval_trilingual_query_relevance_dev_cosine_recall": 0.9244201198853271,
285
+ "step": 1200
286
+ },
287
+ {
288
+ "epoch": 0.0424614580611445,
289
+ "grad_norm": 0.0,
290
+ "learning_rate": 1.061397779229262e-05,
291
+ "loss": 0.0121,
292
+ "step": 1300
293
+ },
294
+ {
295
+ "epoch": 0.0424614580611445,
296
+ "eval_loss": 0.01010512188076973,
297
+ "eval_runtime": 210.034,
298
+ "eval_samples_per_second": 73.074,
299
+ "eval_steps_per_second": 36.537,
300
+ "eval_trilingual_query_relevance_dev_cosine_accuracy": 0.918816783945791,
301
+ "eval_trilingual_query_relevance_dev_cosine_accuracy_threshold": 0.7169110774993896,
302
+ "eval_trilingual_query_relevance_dev_cosine_ap": 0.9740851173949383,
303
+ "eval_trilingual_query_relevance_dev_cosine_f1": 0.9184672628761668,
304
+ "eval_trilingual_query_relevance_dev_cosine_f1_threshold": 0.7098567485809326,
305
+ "eval_trilingual_query_relevance_dev_cosine_mcc": 0.8372486157867537,
306
+ "eval_trilingual_query_relevance_dev_cosine_precision": 0.9202092871157619,
307
+ "eval_trilingual_query_relevance_dev_cosine_recall": 0.9167318217357311,
308
+ "step": 1300
309
+ },
310
+ {
311
+ "epoch": 0.04572772406584792,
312
+ "grad_norm": 0.11284372210502625,
313
+ "learning_rate": 1.1430437622468976e-05,
314
+ "loss": 0.013,
315
+ "step": 1400
316
+ },
317
+ {
318
+ "epoch": 0.04572772406584792,
319
+ "eval_loss": 0.010898754000663757,
320
+ "eval_runtime": 209.771,
321
+ "eval_samples_per_second": 73.165,
322
+ "eval_steps_per_second": 36.583,
323
+ "eval_trilingual_query_relevance_dev_cosine_accuracy": 0.9161454261141517,
324
+ "eval_trilingual_query_relevance_dev_cosine_accuracy_threshold": 0.6819369196891785,
325
+ "eval_trilingual_query_relevance_dev_cosine_ap": 0.9730695267122584,
326
+ "eval_trilingual_query_relevance_dev_cosine_f1": 0.9150102357524927,
327
+ "eval_trilingual_query_relevance_dev_cosine_f1_threshold": 0.6819369196891785,
328
+ "eval_trilingual_query_relevance_dev_cosine_mcc": 0.832587978992555,
329
+ "eval_trilingual_query_relevance_dev_cosine_precision": 0.927567278082742,
330
+ "eval_trilingual_query_relevance_dev_cosine_recall": 0.9027886369559551,
331
+ "step": 1400
332
+ },
333
+ {
334
+ "epoch": 0.048993990070551344,
335
+ "grad_norm": 0.07064260542392731,
336
+ "learning_rate": 1.224689745264533e-05,
337
+ "loss": 0.0123,
338
+ "step": 1500
339
+ },
340
+ {
341
+ "epoch": 0.048993990070551344,
342
+ "eval_loss": 0.010588725097477436,
343
+ "eval_runtime": 210.023,
344
+ "eval_samples_per_second": 73.078,
345
+ "eval_steps_per_second": 36.539,
346
+ "eval_trilingual_query_relevance_dev_cosine_accuracy": 0.9169924420119885,
347
+ "eval_trilingual_query_relevance_dev_cosine_accuracy_threshold": 0.6868540644645691,
348
+ "eval_trilingual_query_relevance_dev_cosine_ap": 0.9738805154459831,
349
+ "eval_trilingual_query_relevance_dev_cosine_f1": 0.916819012797075,
350
+ "eval_trilingual_query_relevance_dev_cosine_f1_threshold": 0.6845408082008362,
351
+ "eval_trilingual_query_relevance_dev_cosine_mcc": 0.8339921348955691,
352
+ "eval_trilingual_query_relevance_dev_cosine_precision": 0.9187385501177702,
353
+ "eval_trilingual_query_relevance_dev_cosine_recall": 0.9149074798019285,
354
+ "step": 1500
355
+ },
356
+ {
357
+ "epoch": 0.052260256075254766,
358
+ "grad_norm": 0.36876916885375977,
359
+ "learning_rate": 1.3063357282821687e-05,
360
+ "loss": 0.0128,
361
+ "step": 1600
362
+ },
363
+ {
364
+ "epoch": 0.052260256075254766,
365
+ "eval_loss": 0.010131288319826126,
366
+ "eval_runtime": 210.0233,
367
+ "eval_samples_per_second": 73.078,
368
+ "eval_steps_per_second": 36.539,
369
+ "eval_trilingual_query_relevance_dev_cosine_accuracy": 0.9192077143601772,
370
+ "eval_trilingual_query_relevance_dev_cosine_accuracy_threshold": 0.6914563179016113,
371
+ "eval_trilingual_query_relevance_dev_cosine_ap": 0.9761206590784132,
372
+ "eval_trilingual_query_relevance_dev_cosine_f1": 0.9197307094769549,
373
+ "eval_trilingual_query_relevance_dev_cosine_f1_threshold": 0.6914563179016113,
374
+ "eval_trilingual_query_relevance_dev_cosine_mcc": 0.8384866223262938,
375
+ "eval_trilingual_query_relevance_dev_cosine_precision": 0.9138152817082583,
376
+ "eval_trilingual_query_relevance_dev_cosine_recall": 0.9257232212666145,
377
+ "step": 1600
378
+ },
379
+ {
380
+ "epoch": 0.055526522079958195,
381
+ "grad_norm": 0.239050954580307,
382
+ "learning_rate": 1.3879817112998043e-05,
383
+ "loss": 0.0118,
384
+ "step": 1700
385
+ },
386
+ {
387
+ "epoch": 0.055526522079958195,
388
+ "eval_loss": 0.009934144094586372,
389
+ "eval_runtime": 210.1215,
390
+ "eval_samples_per_second": 73.043,
391
+ "eval_steps_per_second": 36.522,
392
+ "eval_trilingual_query_relevance_dev_cosine_accuracy": 0.9217487620536878,
393
+ "eval_trilingual_query_relevance_dev_cosine_accuracy_threshold": 0.7016469240188599,
394
+ "eval_trilingual_query_relevance_dev_cosine_ap": 0.9766940535892453,
395
+ "eval_trilingual_query_relevance_dev_cosine_f1": 0.9212339795195466,
396
+ "eval_trilingual_query_relevance_dev_cosine_f1_threshold": 0.6714333891868591,
397
+ "eval_trilingual_query_relevance_dev_cosine_mcc": 0.8408594794514451,
398
+ "eval_trilingual_query_relevance_dev_cosine_precision": 0.9107347510505539,
399
+ "eval_trilingual_query_relevance_dev_cosine_recall": 0.9319781078967944,
400
+ "step": 1700
401
+ },
402
+ {
403
+ "epoch": 0.05879278808466162,
404
+ "grad_norm": 0.04047306999564171,
405
+ "learning_rate": 1.4696276943174397e-05,
406
+ "loss": 0.0116,
407
+ "step": 1800
408
+ },
409
+ {
410
+ "epoch": 0.05879278808466162,
411
+ "eval_loss": 0.009669623337686062,
412
+ "eval_runtime": 210.2475,
413
+ "eval_samples_per_second": 73.0,
414
+ "eval_steps_per_second": 36.5,
415
+ "eval_trilingual_query_relevance_dev_cosine_accuracy": 0.9291764399270264,
416
+ "eval_trilingual_query_relevance_dev_cosine_accuracy_threshold": 0.7117798328399658,
417
+ "eval_trilingual_query_relevance_dev_cosine_ap": 0.9804267342657018,
418
+ "eval_trilingual_query_relevance_dev_cosine_f1": 0.9288958090048008,
419
+ "eval_trilingual_query_relevance_dev_cosine_f1_threshold": 0.6951316595077515,
420
+ "eval_trilingual_query_relevance_dev_cosine_mcc": 0.857211792347683,
421
+ "eval_trilingual_query_relevance_dev_cosine_precision": 0.9249354005167959,
422
+ "eval_trilingual_query_relevance_dev_cosine_recall": 0.9328902788636956,
423
+ "step": 1800
424
+ },
425
+ {
426
+ "epoch": 0.06205905408936504,
427
+ "grad_norm": 0.5182368755340576,
428
+ "learning_rate": 1.5512736773350754e-05,
429
+ "loss": 0.0091,
430
+ "step": 1900
431
+ },
432
+ {
433
+ "epoch": 0.06205905408936504,
434
+ "eval_loss": 0.009580136276781559,
435
+ "eval_runtime": 209.6038,
436
+ "eval_samples_per_second": 73.224,
437
+ "eval_steps_per_second": 36.612,
438
+ "eval_trilingual_query_relevance_dev_cosine_accuracy": 0.9214229867083659,
439
+ "eval_trilingual_query_relevance_dev_cosine_accuracy_threshold": 0.7342114448547363,
440
+ "eval_trilingual_query_relevance_dev_cosine_ap": 0.9762691065142695,
441
+ "eval_trilingual_query_relevance_dev_cosine_f1": 0.9215354586857515,
442
+ "eval_trilingual_query_relevance_dev_cosine_f1_threshold": 0.7342114448547363,
443
+ "eval_trilingual_query_relevance_dev_cosine_mcc": 0.8428494369763666,
444
+ "eval_trilingual_query_relevance_dev_cosine_precision": 0.9202182952182952,
445
+ "eval_trilingual_query_relevance_dev_cosine_recall": 0.9228563982277821,
446
+ "step": 1900
447
+ }
448
+ ],
449
+ "logging_steps": 100,
450
+ "max_steps": 61232,
451
+ "num_input_tokens_seen": 0,
452
+ "num_train_epochs": 2,
453
+ "save_steps": 100,
454
+ "stateful_callbacks": {
455
+ "TrainerControl": {
456
+ "args": {
457
+ "should_epoch_stop": false,
458
+ "should_evaluate": false,
459
+ "should_log": false,
460
+ "should_save": true,
461
+ "should_training_stop": false
462
+ },
463
+ "attributes": {}
464
+ }
465
+ },
466
+ "total_flos": 0.0,
467
+ "train_batch_size": 2,
468
+ "trial_name": null,
469
+ "trial_params": null
470
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:61ed6c66cd6b3f7e9cee8dfc1d57e6b291ee994cf00b23c7d6df5c6bfcf95b1a
3
+ size 5688