alvanli commited on
Commit
e27ff29
1 Parent(s): d7d406e

Added model trained with more data

Browse files
README.md CHANGED
@@ -63,7 +63,7 @@ For training, three datasets were used:
63
  | 0.0546 | 1.33 | 2000 | 0.2887 | 12.119 |
64
  | 0.0293 | 2.01 | 3000 | 0.2727 | 11.646 |
65
  | 0.0214 | 2.66 | 4000 | 0.2741 | 11.760 |
66
- | xx | xx | 5000 | xx | xx |
67
 
68
 
69
  ### Framework versions
 
63
  | 0.0546 | 1.33 | 2000 | 0.2887 | 12.119 |
64
  | 0.0293 | 2.01 | 3000 | 0.2727 | 11.646 |
65
  | 0.0214 | 2.66 | 4000 | 0.2741 | 11.760 |
66
+ | 0.0919 | 3.32 | 5000 | 0.2747 | 11.463 |
67
 
68
 
69
  ### Framework versions
config.json CHANGED
@@ -1,5 +1,5 @@
1
  {
2
- "_name_or_path": "openai/whisper-small",
3
  "activation_dropout": 0.0,
4
  "activation_function": "gelu",
5
  "architectures": [
 
1
  {
2
+ "_name_or_path": "./model_out_trpro_full_canto",
3
  "activation_dropout": 0.0,
4
  "activation_function": "gelu",
5
  "architectures": [
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d9ac03f4d32c5b7758a29ae0a12f70210c956a498e67058c86cbd78ecbb607c6
3
+ size 1934161093
pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f0fc1b0188915501fc1066b2932bcedbe557ab656231371b7ea5278a28d488d6
3
  size 967102601
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:205a0370f20a92cda895469c286011a8d8998d306d5ba3959c971197b6a2d4bb
3
  size 967102601
rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f4e1a74104f1f012cc2ecbe9819a318f3c008d1bd2e782a3cd2eaa6d95a6f367
3
+ size 14511
runs/Dec07_00-02-22_a5b4b6dedeb2/1670371344.7013464/events.out.tfevents.1670371344.a5b4b6dedeb2.15.1 DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:781e6a9003a8d9d24a46ce469705a31a0ddc953f8de2a140e65441a5b97bfcdc
3
- size 5906
 
 
 
 
runs/Dec07_00-02-22_a5b4b6dedeb2/events.out.tfevents.1670371344.a5b4b6dedeb2.15.0 DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:7f092c2f8fd8d703b90fa37b32d7d86bc3df91f844adeae92398dbab75ba9b60
3
- size 4864
 
 
 
 
runs/Dec07_03-26-07_a5b4b6dedeb2/1670383569.2554996/events.out.tfevents.1670383569.a5b4b6dedeb2.80458.1 DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:93dd42dc0b325cd12c6930d1ec50cba42f548ea6ad824a3128336ea016106e1d
3
- size 5906
 
 
 
 
runs/Dec07_03-26-07_a5b4b6dedeb2/events.out.tfevents.1670383569.a5b4b6dedeb2.80458.0 DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:added0e9ad8580e202acf95691e3d20a3bdf9b1df77c8adc4b5b834c2deb3567
3
- size 10667
 
 
 
 
runs/Dec07_13-42-04_a5b4b6dedeb2/1670420526.9102201/events.out.tfevents.1670420526.a5b4b6dedeb2.207425.1 DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:49aef4015485c5766d5a879c6e61911501a712873de0698784acedc67dc64bac
3
- size 5906
 
 
 
 
runs/Dec07_13-42-04_a5b4b6dedeb2/events.out.tfevents.1670420526.a5b4b6dedeb2.207425.0 DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:502296505ce6ef4840105a124f22b0c50d66e738c865f1518504fe018dfe1ad3
3
- size 37731
 
 
 
 
runs/{Dec06_16-11-14_6dc52a944789/1670343076.026156/events.out.tfevents.1670343076.6dc52a944789.2489167.1 → Dec12_00-47-25_b23159305be2/1670806047.5609205/events.out.tfevents.1670806047.b23159305be2.526.1} RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:963207f80131d92324877065dacae1858bbbb066d41985d7dc1ef7cc3a776696
3
- size 5906
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5bcb2c0385ecea395ee5211a1fa2c767f5e7c0826efea079bdb0fcec52779e89
3
+ size 5801
runs/Dec12_00-47-25_b23159305be2/events.out.tfevents.1670806047.b23159305be2.526.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8020afb61ba6589aee28c7b0f129a53ce340d690cd3315dd274063850c117841
3
+ size 37634
runs/Dec06_16-11-14_6dc52a944789/events.out.tfevents.1670343076.6dc52a944789.2489167.0 → scaler.pt RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:196961fa98924b827e898b67a18ffd9f4e641fdc848cbb5e3153643fde496d31
3
- size 8312
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:daab59ab759c6af5bf94e4c81bf87c0ccd19625eadd7be1b4c85cbd0ce45e7a4
3
+ size 557
runs/Dec06_17-51-28_ade490004456/1670349091.6596131/events.out.tfevents.1670349091.ade490004456.135.1 → scheduler.pt RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e6972fbe09408685e93a0788d0aae978d5a5529a2dbf15b01bb6edb7c93f35b2
3
- size 5906
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:42644c5c1e654cf2ffd124d4abf9e0b2ad81fe8a38fb21d55dee1e3ec6bcfe1b
3
+ size 627
tokenizer_config.json CHANGED
@@ -19,7 +19,7 @@
19
  },
20
  "errors": "replace",
21
  "model_max_length": 1024,
22
- "name_or_path": "openai/whisper-small",
23
  "pad_token": null,
24
  "processor_class": "WhisperProcessor",
25
  "return_attention_mask": false,
 
19
  },
20
  "errors": "replace",
21
  "model_max_length": 1024,
22
+ "name_or_path": "./model_out_trpro_full_canto",
23
  "pad_token": null,
24
  "processor_class": "WhisperProcessor",
25
  "return_attention_mask": false,
trainer_state.json ADDED
@@ -0,0 +1,1261 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 11.462658915223917,
3
+ "best_model_checkpoint": "./model_out_trpro_full_canto_resume/checkpoint-5000",
4
+ "epoch": 3.324468085106383,
5
+ "global_step": 5000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.02,
12
+ "learning_rate": 4.4e-07,
13
+ "loss": 2.6129,
14
+ "step": 25
15
+ },
16
+ {
17
+ "epoch": 0.03,
18
+ "learning_rate": 9.400000000000001e-07,
19
+ "loss": 1.9202,
20
+ "step": 50
21
+ },
22
+ {
23
+ "epoch": 0.05,
24
+ "learning_rate": 1.44e-06,
25
+ "loss": 1.0182,
26
+ "step": 75
27
+ },
28
+ {
29
+ "epoch": 0.07,
30
+ "learning_rate": 1.94e-06,
31
+ "loss": 0.3318,
32
+ "step": 100
33
+ },
34
+ {
35
+ "epoch": 0.08,
36
+ "learning_rate": 2.4400000000000004e-06,
37
+ "loss": 0.262,
38
+ "step": 125
39
+ },
40
+ {
41
+ "epoch": 0.1,
42
+ "learning_rate": 2.9400000000000002e-06,
43
+ "loss": 0.2435,
44
+ "step": 150
45
+ },
46
+ {
47
+ "epoch": 0.12,
48
+ "learning_rate": 3.44e-06,
49
+ "loss": 0.2288,
50
+ "step": 175
51
+ },
52
+ {
53
+ "epoch": 0.13,
54
+ "learning_rate": 3.94e-06,
55
+ "loss": 0.2156,
56
+ "step": 200
57
+ },
58
+ {
59
+ "epoch": 0.15,
60
+ "learning_rate": 4.440000000000001e-06,
61
+ "loss": 0.2003,
62
+ "step": 225
63
+ },
64
+ {
65
+ "epoch": 0.17,
66
+ "learning_rate": 4.94e-06,
67
+ "loss": 0.1971,
68
+ "step": 250
69
+ },
70
+ {
71
+ "epoch": 0.18,
72
+ "learning_rate": 5.4400000000000004e-06,
73
+ "loss": 0.1893,
74
+ "step": 275
75
+ },
76
+ {
77
+ "epoch": 0.2,
78
+ "learning_rate": 5.94e-06,
79
+ "loss": 0.1953,
80
+ "step": 300
81
+ },
82
+ {
83
+ "epoch": 0.22,
84
+ "learning_rate": 6.440000000000001e-06,
85
+ "loss": 0.1757,
86
+ "step": 325
87
+ },
88
+ {
89
+ "epoch": 0.23,
90
+ "learning_rate": 6.9400000000000005e-06,
91
+ "loss": 0.1817,
92
+ "step": 350
93
+ },
94
+ {
95
+ "epoch": 0.25,
96
+ "learning_rate": 7.440000000000001e-06,
97
+ "loss": 0.1814,
98
+ "step": 375
99
+ },
100
+ {
101
+ "epoch": 0.27,
102
+ "learning_rate": 7.94e-06,
103
+ "loss": 0.1736,
104
+ "step": 400
105
+ },
106
+ {
107
+ "epoch": 0.28,
108
+ "learning_rate": 8.44e-06,
109
+ "loss": 0.1731,
110
+ "step": 425
111
+ },
112
+ {
113
+ "epoch": 0.3,
114
+ "learning_rate": 8.94e-06,
115
+ "loss": 0.1741,
116
+ "step": 450
117
+ },
118
+ {
119
+ "epoch": 0.32,
120
+ "learning_rate": 9.440000000000001e-06,
121
+ "loss": 0.1611,
122
+ "step": 475
123
+ },
124
+ {
125
+ "epoch": 0.33,
126
+ "learning_rate": 9.940000000000001e-06,
127
+ "loss": 0.152,
128
+ "step": 500
129
+ },
130
+ {
131
+ "epoch": 0.35,
132
+ "learning_rate": 9.951111111111111e-06,
133
+ "loss": 0.1447,
134
+ "step": 525
135
+ },
136
+ {
137
+ "epoch": 0.37,
138
+ "learning_rate": 9.895555555555557e-06,
139
+ "loss": 0.1462,
140
+ "step": 550
141
+ },
142
+ {
143
+ "epoch": 0.38,
144
+ "learning_rate": 9.84e-06,
145
+ "loss": 0.1554,
146
+ "step": 575
147
+ },
148
+ {
149
+ "epoch": 0.4,
150
+ "learning_rate": 9.784444444444445e-06,
151
+ "loss": 0.1528,
152
+ "step": 600
153
+ },
154
+ {
155
+ "epoch": 0.42,
156
+ "learning_rate": 9.72888888888889e-06,
157
+ "loss": 0.1443,
158
+ "step": 625
159
+ },
160
+ {
161
+ "epoch": 0.43,
162
+ "learning_rate": 9.673333333333334e-06,
163
+ "loss": 0.1444,
164
+ "step": 650
165
+ },
166
+ {
167
+ "epoch": 0.45,
168
+ "learning_rate": 9.617777777777778e-06,
169
+ "loss": 0.1436,
170
+ "step": 675
171
+ },
172
+ {
173
+ "epoch": 0.47,
174
+ "learning_rate": 9.562222222222223e-06,
175
+ "loss": 0.135,
176
+ "step": 700
177
+ },
178
+ {
179
+ "epoch": 0.48,
180
+ "learning_rate": 9.506666666666667e-06,
181
+ "loss": 0.1317,
182
+ "step": 725
183
+ },
184
+ {
185
+ "epoch": 0.5,
186
+ "learning_rate": 9.451111111111112e-06,
187
+ "loss": 0.1307,
188
+ "step": 750
189
+ },
190
+ {
191
+ "epoch": 0.52,
192
+ "learning_rate": 9.395555555555556e-06,
193
+ "loss": 0.138,
194
+ "step": 775
195
+ },
196
+ {
197
+ "epoch": 0.53,
198
+ "learning_rate": 9.340000000000002e-06,
199
+ "loss": 0.1283,
200
+ "step": 800
201
+ },
202
+ {
203
+ "epoch": 0.55,
204
+ "learning_rate": 9.284444444444444e-06,
205
+ "loss": 0.1287,
206
+ "step": 825
207
+ },
208
+ {
209
+ "epoch": 0.57,
210
+ "learning_rate": 9.22888888888889e-06,
211
+ "loss": 0.127,
212
+ "step": 850
213
+ },
214
+ {
215
+ "epoch": 0.58,
216
+ "learning_rate": 9.173333333333334e-06,
217
+ "loss": 0.1253,
218
+ "step": 875
219
+ },
220
+ {
221
+ "epoch": 0.6,
222
+ "learning_rate": 9.117777777777778e-06,
223
+ "loss": 0.1206,
224
+ "step": 900
225
+ },
226
+ {
227
+ "epoch": 0.62,
228
+ "learning_rate": 9.062222222222224e-06,
229
+ "loss": 0.1271,
230
+ "step": 925
231
+ },
232
+ {
233
+ "epoch": 0.63,
234
+ "learning_rate": 9.006666666666666e-06,
235
+ "loss": 0.1152,
236
+ "step": 950
237
+ },
238
+ {
239
+ "epoch": 0.65,
240
+ "learning_rate": 8.951111111111112e-06,
241
+ "loss": 0.116,
242
+ "step": 975
243
+ },
244
+ {
245
+ "epoch": 0.66,
246
+ "learning_rate": 8.895555555555556e-06,
247
+ "loss": 0.11,
248
+ "step": 1000
249
+ },
250
+ {
251
+ "epoch": 0.66,
252
+ "eval_cer": 14.637769117741009,
253
+ "eval_loss": 0.32944002747535706,
254
+ "eval_runtime": 1229.289,
255
+ "eval_samples_per_second": 4.548,
256
+ "eval_steps_per_second": 0.569,
257
+ "step": 1000
258
+ },
259
+ {
260
+ "epoch": 0.68,
261
+ "learning_rate": 8.84e-06,
262
+ "loss": 0.1105,
263
+ "step": 1025
264
+ },
265
+ {
266
+ "epoch": 0.7,
267
+ "learning_rate": 8.784444444444446e-06,
268
+ "loss": 0.1091,
269
+ "step": 1050
270
+ },
271
+ {
272
+ "epoch": 0.71,
273
+ "learning_rate": 8.72888888888889e-06,
274
+ "loss": 0.1019,
275
+ "step": 1075
276
+ },
277
+ {
278
+ "epoch": 0.73,
279
+ "learning_rate": 8.673333333333334e-06,
280
+ "loss": 0.1122,
281
+ "step": 1100
282
+ },
283
+ {
284
+ "epoch": 0.75,
285
+ "learning_rate": 8.617777777777778e-06,
286
+ "loss": 0.1119,
287
+ "step": 1125
288
+ },
289
+ {
290
+ "epoch": 0.76,
291
+ "learning_rate": 8.562222222222224e-06,
292
+ "loss": 0.109,
293
+ "step": 1150
294
+ },
295
+ {
296
+ "epoch": 0.78,
297
+ "learning_rate": 8.506666666666668e-06,
298
+ "loss": 0.1039,
299
+ "step": 1175
300
+ },
301
+ {
302
+ "epoch": 0.8,
303
+ "learning_rate": 8.451111111111112e-06,
304
+ "loss": 0.1001,
305
+ "step": 1200
306
+ },
307
+ {
308
+ "epoch": 0.81,
309
+ "learning_rate": 8.395555555555557e-06,
310
+ "loss": 0.1075,
311
+ "step": 1225
312
+ },
313
+ {
314
+ "epoch": 0.83,
315
+ "learning_rate": 8.34e-06,
316
+ "loss": 0.0968,
317
+ "step": 1250
318
+ },
319
+ {
320
+ "epoch": 0.85,
321
+ "learning_rate": 8.284444444444446e-06,
322
+ "loss": 0.0993,
323
+ "step": 1275
324
+ },
325
+ {
326
+ "epoch": 0.86,
327
+ "learning_rate": 8.22888888888889e-06,
328
+ "loss": 0.0982,
329
+ "step": 1300
330
+ },
331
+ {
332
+ "epoch": 0.88,
333
+ "learning_rate": 8.173333333333334e-06,
334
+ "loss": 0.0927,
335
+ "step": 1325
336
+ },
337
+ {
338
+ "epoch": 0.9,
339
+ "learning_rate": 8.11777777777778e-06,
340
+ "loss": 0.0963,
341
+ "step": 1350
342
+ },
343
+ {
344
+ "epoch": 0.91,
345
+ "learning_rate": 8.062222222222222e-06,
346
+ "loss": 0.0939,
347
+ "step": 1375
348
+ },
349
+ {
350
+ "epoch": 0.93,
351
+ "learning_rate": 8.006666666666667e-06,
352
+ "loss": 0.0949,
353
+ "step": 1400
354
+ },
355
+ {
356
+ "epoch": 0.95,
357
+ "learning_rate": 7.951111111111111e-06,
358
+ "loss": 0.0963,
359
+ "step": 1425
360
+ },
361
+ {
362
+ "epoch": 0.96,
363
+ "learning_rate": 7.895555555555557e-06,
364
+ "loss": 0.0926,
365
+ "step": 1450
366
+ },
367
+ {
368
+ "epoch": 0.98,
369
+ "learning_rate": 7.840000000000001e-06,
370
+ "loss": 0.0871,
371
+ "step": 1475
372
+ },
373
+ {
374
+ "epoch": 1.0,
375
+ "learning_rate": 7.784444444444445e-06,
376
+ "loss": 0.0896,
377
+ "step": 1500
378
+ },
379
+ {
380
+ "epoch": 1.01,
381
+ "learning_rate": 7.72888888888889e-06,
382
+ "loss": 0.0679,
383
+ "step": 1525
384
+ },
385
+ {
386
+ "epoch": 1.03,
387
+ "learning_rate": 7.673333333333333e-06,
388
+ "loss": 0.0638,
389
+ "step": 1550
390
+ },
391
+ {
392
+ "epoch": 1.05,
393
+ "learning_rate": 7.617777777777778e-06,
394
+ "loss": 0.0603,
395
+ "step": 1575
396
+ },
397
+ {
398
+ "epoch": 1.06,
399
+ "learning_rate": 7.562222222222223e-06,
400
+ "loss": 0.066,
401
+ "step": 1600
402
+ },
403
+ {
404
+ "epoch": 1.08,
405
+ "learning_rate": 7.506666666666668e-06,
406
+ "loss": 0.0616,
407
+ "step": 1625
408
+ },
409
+ {
410
+ "epoch": 1.1,
411
+ "learning_rate": 7.451111111111111e-06,
412
+ "loss": 0.0632,
413
+ "step": 1650
414
+ },
415
+ {
416
+ "epoch": 1.11,
417
+ "learning_rate": 7.395555555555556e-06,
418
+ "loss": 0.0592,
419
+ "step": 1675
420
+ },
421
+ {
422
+ "epoch": 1.13,
423
+ "learning_rate": 7.340000000000001e-06,
424
+ "loss": 0.0634,
425
+ "step": 1700
426
+ },
427
+ {
428
+ "epoch": 1.15,
429
+ "learning_rate": 7.284444444444445e-06,
430
+ "loss": 0.0587,
431
+ "step": 1725
432
+ },
433
+ {
434
+ "epoch": 1.16,
435
+ "learning_rate": 7.22888888888889e-06,
436
+ "loss": 0.0593,
437
+ "step": 1750
438
+ },
439
+ {
440
+ "epoch": 1.18,
441
+ "learning_rate": 7.173333333333335e-06,
442
+ "loss": 0.0557,
443
+ "step": 1775
444
+ },
445
+ {
446
+ "epoch": 1.2,
447
+ "learning_rate": 7.117777777777778e-06,
448
+ "loss": 0.0602,
449
+ "step": 1800
450
+ },
451
+ {
452
+ "epoch": 1.21,
453
+ "learning_rate": 7.062222222222223e-06,
454
+ "loss": 0.0625,
455
+ "step": 1825
456
+ },
457
+ {
458
+ "epoch": 1.23,
459
+ "learning_rate": 7.006666666666667e-06,
460
+ "loss": 0.0586,
461
+ "step": 1850
462
+ },
463
+ {
464
+ "epoch": 1.25,
465
+ "learning_rate": 6.951111111111112e-06,
466
+ "loss": 0.0538,
467
+ "step": 1875
468
+ },
469
+ {
470
+ "epoch": 1.26,
471
+ "learning_rate": 6.8955555555555565e-06,
472
+ "loss": 0.0558,
473
+ "step": 1900
474
+ },
475
+ {
476
+ "epoch": 1.28,
477
+ "learning_rate": 6.8400000000000014e-06,
478
+ "loss": 0.059,
479
+ "step": 1925
480
+ },
481
+ {
482
+ "epoch": 1.3,
483
+ "learning_rate": 6.784444444444445e-06,
484
+ "loss": 0.0568,
485
+ "step": 1950
486
+ },
487
+ {
488
+ "epoch": 1.31,
489
+ "learning_rate": 6.7288888888888895e-06,
490
+ "loss": 0.0557,
491
+ "step": 1975
492
+ },
493
+ {
494
+ "epoch": 1.33,
495
+ "learning_rate": 6.6733333333333335e-06,
496
+ "loss": 0.0563,
497
+ "step": 2000
498
+ },
499
+ {
500
+ "epoch": 1.33,
501
+ "eval_cer": 12.119082603973679,
502
+ "eval_loss": 0.288747102022171,
503
+ "eval_runtime": 1184.6421,
504
+ "eval_samples_per_second": 4.72,
505
+ "eval_steps_per_second": 0.59,
506
+ "step": 2000
507
+ },
508
+ {
509
+ "epoch": 1.35,
510
+ "learning_rate": 6.617777777777778e-06,
511
+ "loss": 0.0546,
512
+ "step": 2025
513
+ },
514
+ {
515
+ "epoch": 1.36,
516
+ "learning_rate": 6.562222222222223e-06,
517
+ "loss": 0.0537,
518
+ "step": 2050
519
+ },
520
+ {
521
+ "epoch": 1.38,
522
+ "learning_rate": 6.5066666666666665e-06,
523
+ "loss": 0.0535,
524
+ "step": 2075
525
+ },
526
+ {
527
+ "epoch": 1.4,
528
+ "learning_rate": 6.451111111111111e-06,
529
+ "loss": 0.0526,
530
+ "step": 2100
531
+ },
532
+ {
533
+ "epoch": 1.41,
534
+ "learning_rate": 6.395555555555556e-06,
535
+ "loss": 0.0542,
536
+ "step": 2125
537
+ },
538
+ {
539
+ "epoch": 1.43,
540
+ "learning_rate": 6.34e-06,
541
+ "loss": 0.0554,
542
+ "step": 2150
543
+ },
544
+ {
545
+ "epoch": 1.45,
546
+ "learning_rate": 6.284444444444445e-06,
547
+ "loss": 0.0547,
548
+ "step": 2175
549
+ },
550
+ {
551
+ "epoch": 1.46,
552
+ "learning_rate": 6.22888888888889e-06,
553
+ "loss": 0.0549,
554
+ "step": 2200
555
+ },
556
+ {
557
+ "epoch": 1.48,
558
+ "learning_rate": 6.173333333333333e-06,
559
+ "loss": 0.0554,
560
+ "step": 2225
561
+ },
562
+ {
563
+ "epoch": 1.5,
564
+ "learning_rate": 6.117777777777778e-06,
565
+ "loss": 0.0527,
566
+ "step": 2250
567
+ },
568
+ {
569
+ "epoch": 1.51,
570
+ "learning_rate": 6.062222222222223e-06,
571
+ "loss": 0.051,
572
+ "step": 2275
573
+ },
574
+ {
575
+ "epoch": 1.53,
576
+ "learning_rate": 6.006666666666667e-06,
577
+ "loss": 0.0515,
578
+ "step": 2300
579
+ },
580
+ {
581
+ "epoch": 1.55,
582
+ "learning_rate": 5.951111111111112e-06,
583
+ "loss": 0.0524,
584
+ "step": 2325
585
+ },
586
+ {
587
+ "epoch": 1.56,
588
+ "learning_rate": 5.895555555555557e-06,
589
+ "loss": 0.0535,
590
+ "step": 2350
591
+ },
592
+ {
593
+ "epoch": 1.58,
594
+ "learning_rate": 5.84e-06,
595
+ "loss": 0.047,
596
+ "step": 2375
597
+ },
598
+ {
599
+ "epoch": 1.6,
600
+ "learning_rate": 5.784444444444445e-06,
601
+ "loss": 0.0524,
602
+ "step": 2400
603
+ },
604
+ {
605
+ "epoch": 1.61,
606
+ "learning_rate": 5.72888888888889e-06,
607
+ "loss": 0.0447,
608
+ "step": 2425
609
+ },
610
+ {
611
+ "epoch": 1.63,
612
+ "learning_rate": 5.673333333333334e-06,
613
+ "loss": 0.0485,
614
+ "step": 2450
615
+ },
616
+ {
617
+ "epoch": 1.65,
618
+ "learning_rate": 5.617777777777779e-06,
619
+ "loss": 0.0497,
620
+ "step": 2475
621
+ },
622
+ {
623
+ "epoch": 1.66,
624
+ "learning_rate": 5.562222222222222e-06,
625
+ "loss": 0.046,
626
+ "step": 2500
627
+ },
628
+ {
629
+ "epoch": 1.68,
630
+ "learning_rate": 5.506666666666667e-06,
631
+ "loss": 0.0526,
632
+ "step": 2525
633
+ },
634
+ {
635
+ "epoch": 1.7,
636
+ "learning_rate": 5.451111111111112e-06,
637
+ "loss": 0.0465,
638
+ "step": 2550
639
+ },
640
+ {
641
+ "epoch": 1.71,
642
+ "learning_rate": 5.3955555555555565e-06,
643
+ "loss": 0.0468,
644
+ "step": 2575
645
+ },
646
+ {
647
+ "epoch": 1.73,
648
+ "learning_rate": 5.3400000000000005e-06,
649
+ "loss": 0.0492,
650
+ "step": 2600
651
+ },
652
+ {
653
+ "epoch": 1.75,
654
+ "learning_rate": 5.2844444444444454e-06,
655
+ "loss": 0.0475,
656
+ "step": 2625
657
+ },
658
+ {
659
+ "epoch": 1.76,
660
+ "learning_rate": 5.228888888888889e-06,
661
+ "loss": 0.0463,
662
+ "step": 2650
663
+ },
664
+ {
665
+ "epoch": 1.78,
666
+ "learning_rate": 5.1733333333333335e-06,
667
+ "loss": 0.0462,
668
+ "step": 2675
669
+ },
670
+ {
671
+ "epoch": 1.8,
672
+ "learning_rate": 5.117777777777778e-06,
673
+ "loss": 0.0505,
674
+ "step": 2700
675
+ },
676
+ {
677
+ "epoch": 1.81,
678
+ "learning_rate": 5.062222222222222e-06,
679
+ "loss": 0.0467,
680
+ "step": 2725
681
+ },
682
+ {
683
+ "epoch": 1.83,
684
+ "learning_rate": 5.006666666666667e-06,
685
+ "loss": 0.0454,
686
+ "step": 2750
687
+ },
688
+ {
689
+ "epoch": 1.85,
690
+ "learning_rate": 4.951111111111111e-06,
691
+ "loss": 0.0448,
692
+ "step": 2775
693
+ },
694
+ {
695
+ "epoch": 1.86,
696
+ "learning_rate": 4.895555555555556e-06,
697
+ "loss": 0.0442,
698
+ "step": 2800
699
+ },
700
+ {
701
+ "epoch": 1.88,
702
+ "learning_rate": 4.84e-06,
703
+ "loss": 0.0419,
704
+ "step": 2825
705
+ },
706
+ {
707
+ "epoch": 1.89,
708
+ "learning_rate": 4.784444444444445e-06,
709
+ "loss": 0.0441,
710
+ "step": 2850
711
+ },
712
+ {
713
+ "epoch": 1.91,
714
+ "learning_rate": 4.728888888888889e-06,
715
+ "loss": 0.0474,
716
+ "step": 2875
717
+ },
718
+ {
719
+ "epoch": 1.93,
720
+ "learning_rate": 4.673333333333333e-06,
721
+ "loss": 0.0428,
722
+ "step": 2900
723
+ },
724
+ {
725
+ "epoch": 1.94,
726
+ "learning_rate": 4.617777777777778e-06,
727
+ "loss": 0.0399,
728
+ "step": 2925
729
+ },
730
+ {
731
+ "epoch": 1.96,
732
+ "learning_rate": 4.562222222222222e-06,
733
+ "loss": 0.0388,
734
+ "step": 2950
735
+ },
736
+ {
737
+ "epoch": 1.98,
738
+ "learning_rate": 4.506666666666667e-06,
739
+ "loss": 0.0419,
740
+ "step": 2975
741
+ },
742
+ {
743
+ "epoch": 1.99,
744
+ "learning_rate": 4.451111111111112e-06,
745
+ "loss": 0.0435,
746
+ "step": 3000
747
+ },
748
+ {
749
+ "epoch": 1.99,
750
+ "eval_cer": 11.646329777039545,
751
+ "eval_loss": 0.2727377712726593,
752
+ "eval_runtime": 1258.956,
753
+ "eval_samples_per_second": 4.441,
754
+ "eval_steps_per_second": 0.555,
755
+ "step": 3000
756
+ },
757
+ {
758
+ "epoch": 2.01,
759
+ "learning_rate": 4.395555555555556e-06,
760
+ "loss": 0.0293,
761
+ "step": 3025
762
+ },
763
+ {
764
+ "epoch": 2.03,
765
+ "learning_rate": 4.34e-06,
766
+ "loss": 0.0234,
767
+ "step": 3050
768
+ },
769
+ {
770
+ "epoch": 2.04,
771
+ "learning_rate": 4.284444444444445e-06,
772
+ "loss": 0.0246,
773
+ "step": 3075
774
+ },
775
+ {
776
+ "epoch": 2.06,
777
+ "learning_rate": 4.228888888888889e-06,
778
+ "loss": 0.0241,
779
+ "step": 3100
780
+ },
781
+ {
782
+ "epoch": 2.08,
783
+ "learning_rate": 4.173333333333334e-06,
784
+ "loss": 0.0244,
785
+ "step": 3125
786
+ },
787
+ {
788
+ "epoch": 2.09,
789
+ "learning_rate": 4.117777777777779e-06,
790
+ "loss": 0.0239,
791
+ "step": 3150
792
+ },
793
+ {
794
+ "epoch": 2.11,
795
+ "learning_rate": 4.062222222222223e-06,
796
+ "loss": 0.0237,
797
+ "step": 3175
798
+ },
799
+ {
800
+ "epoch": 2.13,
801
+ "learning_rate": 4.006666666666667e-06,
802
+ "loss": 0.0229,
803
+ "step": 3200
804
+ },
805
+ {
806
+ "epoch": 2.14,
807
+ "learning_rate": 3.951111111111112e-06,
808
+ "loss": 0.0228,
809
+ "step": 3225
810
+ },
811
+ {
812
+ "epoch": 2.16,
813
+ "learning_rate": 3.895555555555556e-06,
814
+ "loss": 0.0232,
815
+ "step": 3250
816
+ },
817
+ {
818
+ "epoch": 2.18,
819
+ "learning_rate": 3.8400000000000005e-06,
820
+ "loss": 0.0248,
821
+ "step": 3275
822
+ },
823
+ {
824
+ "epoch": 2.19,
825
+ "learning_rate": 3.784444444444445e-06,
826
+ "loss": 0.0214,
827
+ "step": 3300
828
+ },
829
+ {
830
+ "epoch": 2.21,
831
+ "learning_rate": 3.728888888888889e-06,
832
+ "loss": 0.0233,
833
+ "step": 3325
834
+ },
835
+ {
836
+ "epoch": 2.23,
837
+ "learning_rate": 3.673333333333334e-06,
838
+ "loss": 0.0231,
839
+ "step": 3350
840
+ },
841
+ {
842
+ "epoch": 2.24,
843
+ "learning_rate": 3.617777777777778e-06,
844
+ "loss": 0.0228,
845
+ "step": 3375
846
+ },
847
+ {
848
+ "epoch": 2.26,
849
+ "learning_rate": 3.5622222222222224e-06,
850
+ "loss": 0.0236,
851
+ "step": 3400
852
+ },
853
+ {
854
+ "epoch": 2.28,
855
+ "learning_rate": 3.5066666666666673e-06,
856
+ "loss": 0.0234,
857
+ "step": 3425
858
+ },
859
+ {
860
+ "epoch": 2.29,
861
+ "learning_rate": 3.4511111111111113e-06,
862
+ "loss": 0.0233,
863
+ "step": 3450
864
+ },
865
+ {
866
+ "epoch": 2.31,
867
+ "learning_rate": 3.3955555555555558e-06,
868
+ "loss": 0.0231,
869
+ "step": 3475
870
+ },
871
+ {
872
+ "epoch": 2.33,
873
+ "learning_rate": 3.3400000000000006e-06,
874
+ "loss": 0.023,
875
+ "step": 3500
876
+ },
877
+ {
878
+ "epoch": 2.34,
879
+ "learning_rate": 3.2844444444444447e-06,
880
+ "loss": 0.023,
881
+ "step": 3525
882
+ },
883
+ {
884
+ "epoch": 2.36,
885
+ "learning_rate": 3.228888888888889e-06,
886
+ "loss": 0.0218,
887
+ "step": 3550
888
+ },
889
+ {
890
+ "epoch": 2.38,
891
+ "learning_rate": 3.173333333333334e-06,
892
+ "loss": 0.0212,
893
+ "step": 3575
894
+ },
895
+ {
896
+ "epoch": 2.39,
897
+ "learning_rate": 3.117777777777778e-06,
898
+ "loss": 0.0227,
899
+ "step": 3600
900
+ },
901
+ {
902
+ "epoch": 2.41,
903
+ "learning_rate": 3.0622222222222225e-06,
904
+ "loss": 0.0217,
905
+ "step": 3625
906
+ },
907
+ {
908
+ "epoch": 2.43,
909
+ "learning_rate": 3.0066666666666674e-06,
910
+ "loss": 0.0225,
911
+ "step": 3650
912
+ },
913
+ {
914
+ "epoch": 2.44,
915
+ "learning_rate": 2.9511111111111114e-06,
916
+ "loss": 0.0196,
917
+ "step": 3675
918
+ },
919
+ {
920
+ "epoch": 2.46,
921
+ "learning_rate": 2.895555555555556e-06,
922
+ "loss": 0.0225,
923
+ "step": 3700
924
+ },
925
+ {
926
+ "epoch": 2.48,
927
+ "learning_rate": 2.84e-06,
928
+ "loss": 0.024,
929
+ "step": 3725
930
+ },
931
+ {
932
+ "epoch": 2.49,
933
+ "learning_rate": 2.784444444444445e-06,
934
+ "loss": 0.0221,
935
+ "step": 3750
936
+ },
937
+ {
938
+ "epoch": 2.51,
939
+ "learning_rate": 2.7288888888888893e-06,
940
+ "loss": 0.0202,
941
+ "step": 3775
942
+ },
943
+ {
944
+ "epoch": 2.53,
945
+ "learning_rate": 2.6733333333333333e-06,
946
+ "loss": 0.0217,
947
+ "step": 3800
948
+ },
949
+ {
950
+ "epoch": 2.54,
951
+ "learning_rate": 2.617777777777778e-06,
952
+ "loss": 0.0199,
953
+ "step": 3825
954
+ },
955
+ {
956
+ "epoch": 2.56,
957
+ "learning_rate": 2.5622222222222226e-06,
958
+ "loss": 0.023,
959
+ "step": 3850
960
+ },
961
+ {
962
+ "epoch": 2.58,
963
+ "learning_rate": 2.5066666666666667e-06,
964
+ "loss": 0.0236,
965
+ "step": 3875
966
+ },
967
+ {
968
+ "epoch": 2.59,
969
+ "learning_rate": 2.451111111111111e-06,
970
+ "loss": 0.02,
971
+ "step": 3900
972
+ },
973
+ {
974
+ "epoch": 2.61,
975
+ "learning_rate": 2.3955555555555556e-06,
976
+ "loss": 0.0225,
977
+ "step": 3925
978
+ },
979
+ {
980
+ "epoch": 2.63,
981
+ "learning_rate": 2.3400000000000005e-06,
982
+ "loss": 0.0212,
983
+ "step": 3950
984
+ },
985
+ {
986
+ "epoch": 2.64,
987
+ "learning_rate": 2.2844444444444445e-06,
988
+ "loss": 0.0208,
989
+ "step": 3975
990
+ },
991
+ {
992
+ "epoch": 2.66,
993
+ "learning_rate": 2.228888888888889e-06,
994
+ "loss": 0.0214,
995
+ "step": 4000
996
+ },
997
+ {
998
+ "epoch": 2.66,
999
+ "eval_cer": 11.759726569986583,
1000
+ "eval_loss": 0.27413761615753174,
1001
+ "eval_runtime": 1247.8057,
1002
+ "eval_samples_per_second": 4.481,
1003
+ "eval_steps_per_second": 0.56,
1004
+ "step": 4000
1005
+ },
1006
+ {
1007
+ "epoch": 2.68,
1008
+ "learning_rate": 2.1733333333333334e-06,
1009
+ "loss": 0.0203,
1010
+ "step": 4025
1011
+ },
1012
+ {
1013
+ "epoch": 2.69,
1014
+ "learning_rate": 2.117777777777778e-06,
1015
+ "loss": 0.0198,
1016
+ "step": 4050
1017
+ },
1018
+ {
1019
+ "epoch": 2.71,
1020
+ "learning_rate": 2.0622222222222223e-06,
1021
+ "loss": 0.0204,
1022
+ "step": 4075
1023
+ },
1024
+ {
1025
+ "epoch": 2.73,
1026
+ "learning_rate": 2.006666666666667e-06,
1027
+ "loss": 0.0194,
1028
+ "step": 4100
1029
+ },
1030
+ {
1031
+ "epoch": 2.74,
1032
+ "learning_rate": 1.9511111111111113e-06,
1033
+ "loss": 0.021,
1034
+ "step": 4125
1035
+ },
1036
+ {
1037
+ "epoch": 2.76,
1038
+ "learning_rate": 1.8955555555555557e-06,
1039
+ "loss": 0.0198,
1040
+ "step": 4150
1041
+ },
1042
+ {
1043
+ "epoch": 2.78,
1044
+ "learning_rate": 1.8400000000000002e-06,
1045
+ "loss": 0.021,
1046
+ "step": 4175
1047
+ },
1048
+ {
1049
+ "epoch": 2.79,
1050
+ "learning_rate": 1.7844444444444444e-06,
1051
+ "loss": 0.0203,
1052
+ "step": 4200
1053
+ },
1054
+ {
1055
+ "epoch": 2.81,
1056
+ "learning_rate": 1.728888888888889e-06,
1057
+ "loss": 0.0225,
1058
+ "step": 4225
1059
+ },
1060
+ {
1061
+ "epoch": 2.83,
1062
+ "learning_rate": 1.6733333333333335e-06,
1063
+ "loss": 0.0198,
1064
+ "step": 4250
1065
+ },
1066
+ {
1067
+ "epoch": 2.84,
1068
+ "learning_rate": 1.6177777777777778e-06,
1069
+ "loss": 0.0214,
1070
+ "step": 4275
1071
+ },
1072
+ {
1073
+ "epoch": 2.86,
1074
+ "learning_rate": 1.5622222222222225e-06,
1075
+ "loss": 0.0209,
1076
+ "step": 4300
1077
+ },
1078
+ {
1079
+ "epoch": 2.88,
1080
+ "learning_rate": 1.506666666666667e-06,
1081
+ "loss": 0.0187,
1082
+ "step": 4325
1083
+ },
1084
+ {
1085
+ "epoch": 2.89,
1086
+ "learning_rate": 1.4511111111111112e-06,
1087
+ "loss": 0.0188,
1088
+ "step": 4350
1089
+ },
1090
+ {
1091
+ "epoch": 2.91,
1092
+ "learning_rate": 1.3955555555555556e-06,
1093
+ "loss": 0.0196,
1094
+ "step": 4375
1095
+ },
1096
+ {
1097
+ "epoch": 2.93,
1098
+ "learning_rate": 1.34e-06,
1099
+ "loss": 0.0209,
1100
+ "step": 4400
1101
+ },
1102
+ {
1103
+ "epoch": 2.94,
1104
+ "learning_rate": 1.2844444444444445e-06,
1105
+ "loss": 0.0206,
1106
+ "step": 4425
1107
+ },
1108
+ {
1109
+ "epoch": 2.96,
1110
+ "learning_rate": 1.228888888888889e-06,
1111
+ "loss": 0.0166,
1112
+ "step": 4450
1113
+ },
1114
+ {
1115
+ "epoch": 2.98,
1116
+ "learning_rate": 1.1733333333333335e-06,
1117
+ "loss": 0.0195,
1118
+ "step": 4475
1119
+ },
1120
+ {
1121
+ "epoch": 2.99,
1122
+ "learning_rate": 1.117777777777778e-06,
1123
+ "loss": 0.0194,
1124
+ "step": 4500
1125
+ },
1126
+ {
1127
+ "epoch": 3.01,
1128
+ "learning_rate": 1.0622222222222222e-06,
1129
+ "loss": 0.0166,
1130
+ "step": 4525
1131
+ },
1132
+ {
1133
+ "epoch": 3.03,
1134
+ "learning_rate": 1.0066666666666668e-06,
1135
+ "loss": 0.0154,
1136
+ "step": 4550
1137
+ },
1138
+ {
1139
+ "epoch": 3.04,
1140
+ "learning_rate": 9.511111111111111e-07,
1141
+ "loss": 0.0132,
1142
+ "step": 4575
1143
+ },
1144
+ {
1145
+ "epoch": 3.06,
1146
+ "learning_rate": 8.955555555555557e-07,
1147
+ "loss": 0.0132,
1148
+ "step": 4600
1149
+ },
1150
+ {
1151
+ "epoch": 3.08,
1152
+ "learning_rate": 8.400000000000001e-07,
1153
+ "loss": 0.0146,
1154
+ "step": 4625
1155
+ },
1156
+ {
1157
+ "epoch": 3.09,
1158
+ "learning_rate": 7.844444444444445e-07,
1159
+ "loss": 0.0132,
1160
+ "step": 4650
1161
+ },
1162
+ {
1163
+ "epoch": 3.11,
1164
+ "learning_rate": 7.28888888888889e-07,
1165
+ "loss": 0.0128,
1166
+ "step": 4675
1167
+ },
1168
+ {
1169
+ "epoch": 3.12,
1170
+ "learning_rate": 6.733333333333334e-07,
1171
+ "loss": 0.0111,
1172
+ "step": 4700
1173
+ },
1174
+ {
1175
+ "epoch": 3.14,
1176
+ "learning_rate": 6.177777777777778e-07,
1177
+ "loss": 0.0132,
1178
+ "step": 4725
1179
+ },
1180
+ {
1181
+ "epoch": 3.16,
1182
+ "learning_rate": 5.622222222222223e-07,
1183
+ "loss": 0.0132,
1184
+ "step": 4750
1185
+ },
1186
+ {
1187
+ "epoch": 3.17,
1188
+ "learning_rate": 5.066666666666667e-07,
1189
+ "loss": 0.0132,
1190
+ "step": 4775
1191
+ },
1192
+ {
1193
+ "epoch": 3.19,
1194
+ "learning_rate": 4.511111111111111e-07,
1195
+ "loss": 0.0132,
1196
+ "step": 4800
1197
+ },
1198
+ {
1199
+ "epoch": 3.21,
1200
+ "learning_rate": 3.9555555555555557e-07,
1201
+ "loss": 0.0127,
1202
+ "step": 4825
1203
+ },
1204
+ {
1205
+ "epoch": 3.22,
1206
+ "learning_rate": 3.4000000000000003e-07,
1207
+ "loss": 0.0132,
1208
+ "step": 4850
1209
+ },
1210
+ {
1211
+ "epoch": 3.24,
1212
+ "learning_rate": 2.844444444444445e-07,
1213
+ "loss": 0.0139,
1214
+ "step": 4875
1215
+ },
1216
+ {
1217
+ "epoch": 3.26,
1218
+ "learning_rate": 2.2888888888888892e-07,
1219
+ "loss": 0.0149,
1220
+ "step": 4900
1221
+ },
1222
+ {
1223
+ "epoch": 3.27,
1224
+ "learning_rate": 1.7333333333333335e-07,
1225
+ "loss": 0.0127,
1226
+ "step": 4925
1227
+ },
1228
+ {
1229
+ "epoch": 3.29,
1230
+ "learning_rate": 1.1777777777777778e-07,
1231
+ "loss": 0.0125,
1232
+ "step": 4950
1233
+ },
1234
+ {
1235
+ "epoch": 3.31,
1236
+ "learning_rate": 6.222222222222223e-08,
1237
+ "loss": 0.0141,
1238
+ "step": 4975
1239
+ },
1240
+ {
1241
+ "epoch": 3.32,
1242
+ "learning_rate": 6.666666666666667e-09,
1243
+ "loss": 0.0126,
1244
+ "step": 5000
1245
+ },
1246
+ {
1247
+ "epoch": 3.32,
1248
+ "eval_cer": 11.462658915223917,
1249
+ "eval_loss": 0.2747412323951721,
1250
+ "eval_runtime": 1231.0075,
1251
+ "eval_samples_per_second": 4.542,
1252
+ "eval_steps_per_second": 0.568,
1253
+ "step": 5000
1254
+ }
1255
+ ],
1256
+ "max_steps": 5000,
1257
+ "num_train_epochs": 4,
1258
+ "total_flos": 1.7312699901837312e+20,
1259
+ "trial_name": null,
1260
+ "trial_params": null
1261
+ }
runs/Dec06_17-51-28_ade490004456/events.out.tfevents.1670349091.ade490004456.135.0 → training_args.bin RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ccd55975cd01dcd948c44d49069608e60cdf9195f74d32b5fbea5878a33ac2c0
3
- size 4402
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ab18e6f164dad4bb92a27e8f17e415043bba52be1ef0016fd54b6b0f699a0abb
3
+ size 3579