End of training
Browse files
all_results.json
ADDED
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 1.0,
|
3 |
+
"eval_loss": 0.13065485656261444,
|
4 |
+
"eval_runtime": 17216.8419,
|
5 |
+
"eval_samples_per_second": 0.934,
|
6 |
+
"eval_steps_per_second": 0.117,
|
7 |
+
"eval_wer": 6.313937972585015,
|
8 |
+
"train_loss": 0.12058185605208079,
|
9 |
+
"train_runtime": 159323.6173,
|
10 |
+
"train_samples_per_second": 1.205,
|
11 |
+
"train_steps_per_second": 0.019
|
12 |
+
}
|
eval_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 1.0,
|
3 |
+
"eval_loss": 0.13065485656261444,
|
4 |
+
"eval_runtime": 17216.8419,
|
5 |
+
"eval_samples_per_second": 0.934,
|
6 |
+
"eval_steps_per_second": 0.117,
|
7 |
+
"eval_wer": 6.313937972585015
|
8 |
+
}
|
runs/Dec12_23-14-58_150-136-118-254/events.out.tfevents.1671063543.150-136-118-254.4000718.2
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1297a7313ba8e1fbab453be0e98e2b77888d73b6fb224eb858ffb75d9cabc45d
|
3 |
+
size 358
|
train_results.json
ADDED
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 1.0,
|
3 |
+
"train_loss": 0.12058185605208079,
|
4 |
+
"train_runtime": 159323.6173,
|
5 |
+
"train_samples_per_second": 1.205,
|
6 |
+
"train_steps_per_second": 0.019
|
7 |
+
}
|
trainer_state.json
ADDED
@@ -0,0 +1,799 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": 6.313937972585015,
|
3 |
+
"best_model_checkpoint": "./checkpoint-3000",
|
4 |
+
"epoch": 1.0,
|
5 |
+
"global_step": 3000,
|
6 |
+
"is_hyper_param_search": false,
|
7 |
+
"is_local_process_zero": true,
|
8 |
+
"is_world_process_zero": true,
|
9 |
+
"log_history": [
|
10 |
+
{
|
11 |
+
"epoch": 0.01,
|
12 |
+
"learning_rate": 4.6000000000000004e-07,
|
13 |
+
"loss": 0.6304,
|
14 |
+
"step": 25
|
15 |
+
},
|
16 |
+
{
|
17 |
+
"epoch": 0.02,
|
18 |
+
"learning_rate": 9.400000000000001e-07,
|
19 |
+
"loss": 0.4908,
|
20 |
+
"step": 50
|
21 |
+
},
|
22 |
+
{
|
23 |
+
"epoch": 0.03,
|
24 |
+
"learning_rate": 1.44e-06,
|
25 |
+
"loss": 0.2708,
|
26 |
+
"step": 75
|
27 |
+
},
|
28 |
+
{
|
29 |
+
"epoch": 0.03,
|
30 |
+
"learning_rate": 1.94e-06,
|
31 |
+
"loss": 0.2194,
|
32 |
+
"step": 100
|
33 |
+
},
|
34 |
+
{
|
35 |
+
"epoch": 0.04,
|
36 |
+
"learning_rate": 2.4400000000000004e-06,
|
37 |
+
"loss": 0.1984,
|
38 |
+
"step": 125
|
39 |
+
},
|
40 |
+
{
|
41 |
+
"epoch": 0.05,
|
42 |
+
"learning_rate": 2.9400000000000002e-06,
|
43 |
+
"loss": 0.1976,
|
44 |
+
"step": 150
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.06,
|
48 |
+
"learning_rate": 3.44e-06,
|
49 |
+
"loss": 0.1758,
|
50 |
+
"step": 175
|
51 |
+
},
|
52 |
+
{
|
53 |
+
"epoch": 0.07,
|
54 |
+
"learning_rate": 3.94e-06,
|
55 |
+
"loss": 0.1747,
|
56 |
+
"step": 200
|
57 |
+
},
|
58 |
+
{
|
59 |
+
"epoch": 0.07,
|
60 |
+
"learning_rate": 4.440000000000001e-06,
|
61 |
+
"loss": 0.1578,
|
62 |
+
"step": 225
|
63 |
+
},
|
64 |
+
{
|
65 |
+
"epoch": 0.08,
|
66 |
+
"learning_rate": 4.94e-06,
|
67 |
+
"loss": 0.1605,
|
68 |
+
"step": 250
|
69 |
+
},
|
70 |
+
{
|
71 |
+
"epoch": 0.09,
|
72 |
+
"learning_rate": 5.4400000000000004e-06,
|
73 |
+
"loss": 0.1654,
|
74 |
+
"step": 275
|
75 |
+
},
|
76 |
+
{
|
77 |
+
"epoch": 0.1,
|
78 |
+
"learning_rate": 5.94e-06,
|
79 |
+
"loss": 0.1633,
|
80 |
+
"step": 300
|
81 |
+
},
|
82 |
+
{
|
83 |
+
"epoch": 0.11,
|
84 |
+
"learning_rate": 6.440000000000001e-06,
|
85 |
+
"loss": 0.1522,
|
86 |
+
"step": 325
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.12,
|
90 |
+
"learning_rate": 6.9400000000000005e-06,
|
91 |
+
"loss": 0.1494,
|
92 |
+
"step": 350
|
93 |
+
},
|
94 |
+
{
|
95 |
+
"epoch": 0.12,
|
96 |
+
"learning_rate": 7.440000000000001e-06,
|
97 |
+
"loss": 0.1372,
|
98 |
+
"step": 375
|
99 |
+
},
|
100 |
+
{
|
101 |
+
"epoch": 0.13,
|
102 |
+
"learning_rate": 7.94e-06,
|
103 |
+
"loss": 0.1338,
|
104 |
+
"step": 400
|
105 |
+
},
|
106 |
+
{
|
107 |
+
"epoch": 0.14,
|
108 |
+
"learning_rate": 8.44e-06,
|
109 |
+
"loss": 0.1441,
|
110 |
+
"step": 425
|
111 |
+
},
|
112 |
+
{
|
113 |
+
"epoch": 0.15,
|
114 |
+
"learning_rate": 8.94e-06,
|
115 |
+
"loss": 0.1401,
|
116 |
+
"step": 450
|
117 |
+
},
|
118 |
+
{
|
119 |
+
"epoch": 0.16,
|
120 |
+
"learning_rate": 9.440000000000001e-06,
|
121 |
+
"loss": 0.1274,
|
122 |
+
"step": 475
|
123 |
+
},
|
124 |
+
{
|
125 |
+
"epoch": 0.17,
|
126 |
+
"learning_rate": 9.940000000000001e-06,
|
127 |
+
"loss": 0.127,
|
128 |
+
"step": 500
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 0.17,
|
132 |
+
"eval_loss": 0.15694677829742432,
|
133 |
+
"eval_runtime": 17256.1308,
|
134 |
+
"eval_samples_per_second": 0.932,
|
135 |
+
"eval_steps_per_second": 0.117,
|
136 |
+
"eval_wer": 7.258771043090319,
|
137 |
+
"step": 500
|
138 |
+
},
|
139 |
+
{
|
140 |
+
"epoch": 0.17,
|
141 |
+
"learning_rate": 9.912000000000001e-06,
|
142 |
+
"loss": 0.1335,
|
143 |
+
"step": 525
|
144 |
+
},
|
145 |
+
{
|
146 |
+
"epoch": 0.18,
|
147 |
+
"learning_rate": 9.812e-06,
|
148 |
+
"loss": 0.1282,
|
149 |
+
"step": 550
|
150 |
+
},
|
151 |
+
{
|
152 |
+
"epoch": 0.19,
|
153 |
+
"learning_rate": 9.716000000000002e-06,
|
154 |
+
"loss": 0.1318,
|
155 |
+
"step": 575
|
156 |
+
},
|
157 |
+
{
|
158 |
+
"epoch": 0.2,
|
159 |
+
"learning_rate": 9.616e-06,
|
160 |
+
"loss": 0.1305,
|
161 |
+
"step": 600
|
162 |
+
},
|
163 |
+
{
|
164 |
+
"epoch": 0.21,
|
165 |
+
"learning_rate": 9.516e-06,
|
166 |
+
"loss": 0.1293,
|
167 |
+
"step": 625
|
168 |
+
},
|
169 |
+
{
|
170 |
+
"epoch": 0.22,
|
171 |
+
"learning_rate": 9.416000000000001e-06,
|
172 |
+
"loss": 0.1148,
|
173 |
+
"step": 650
|
174 |
+
},
|
175 |
+
{
|
176 |
+
"epoch": 0.23,
|
177 |
+
"learning_rate": 9.316e-06,
|
178 |
+
"loss": 0.1349,
|
179 |
+
"step": 675
|
180 |
+
},
|
181 |
+
{
|
182 |
+
"epoch": 0.23,
|
183 |
+
"learning_rate": 9.216000000000001e-06,
|
184 |
+
"loss": 0.1178,
|
185 |
+
"step": 700
|
186 |
+
},
|
187 |
+
{
|
188 |
+
"epoch": 0.24,
|
189 |
+
"learning_rate": 9.116e-06,
|
190 |
+
"loss": 0.1252,
|
191 |
+
"step": 725
|
192 |
+
},
|
193 |
+
{
|
194 |
+
"epoch": 0.25,
|
195 |
+
"learning_rate": 9.016e-06,
|
196 |
+
"loss": 0.1253,
|
197 |
+
"step": 750
|
198 |
+
},
|
199 |
+
{
|
200 |
+
"epoch": 0.26,
|
201 |
+
"learning_rate": 8.916e-06,
|
202 |
+
"loss": 0.1203,
|
203 |
+
"step": 775
|
204 |
+
},
|
205 |
+
{
|
206 |
+
"epoch": 0.27,
|
207 |
+
"learning_rate": 8.816000000000002e-06,
|
208 |
+
"loss": 0.1222,
|
209 |
+
"step": 800
|
210 |
+
},
|
211 |
+
{
|
212 |
+
"epoch": 0.28,
|
213 |
+
"learning_rate": 8.716000000000001e-06,
|
214 |
+
"loss": 0.1207,
|
215 |
+
"step": 825
|
216 |
+
},
|
217 |
+
{
|
218 |
+
"epoch": 0.28,
|
219 |
+
"learning_rate": 8.616000000000002e-06,
|
220 |
+
"loss": 0.1133,
|
221 |
+
"step": 850
|
222 |
+
},
|
223 |
+
{
|
224 |
+
"epoch": 0.29,
|
225 |
+
"learning_rate": 8.516000000000001e-06,
|
226 |
+
"loss": 0.1179,
|
227 |
+
"step": 875
|
228 |
+
},
|
229 |
+
{
|
230 |
+
"epoch": 0.3,
|
231 |
+
"learning_rate": 8.416e-06,
|
232 |
+
"loss": 0.1164,
|
233 |
+
"step": 900
|
234 |
+
},
|
235 |
+
{
|
236 |
+
"epoch": 0.31,
|
237 |
+
"learning_rate": 8.316000000000001e-06,
|
238 |
+
"loss": 0.1176,
|
239 |
+
"step": 925
|
240 |
+
},
|
241 |
+
{
|
242 |
+
"epoch": 0.32,
|
243 |
+
"learning_rate": 8.216e-06,
|
244 |
+
"loss": 0.1151,
|
245 |
+
"step": 950
|
246 |
+
},
|
247 |
+
{
|
248 |
+
"epoch": 0.33,
|
249 |
+
"learning_rate": 8.116e-06,
|
250 |
+
"loss": 0.1205,
|
251 |
+
"step": 975
|
252 |
+
},
|
253 |
+
{
|
254 |
+
"epoch": 0.33,
|
255 |
+
"learning_rate": 8.016e-06,
|
256 |
+
"loss": 0.1116,
|
257 |
+
"step": 1000
|
258 |
+
},
|
259 |
+
{
|
260 |
+
"epoch": 0.33,
|
261 |
+
"eval_loss": 0.15050184726715088,
|
262 |
+
"eval_runtime": 17277.2664,
|
263 |
+
"eval_samples_per_second": 0.931,
|
264 |
+
"eval_steps_per_second": 0.116,
|
265 |
+
"eval_wer": 7.237205666302751,
|
266 |
+
"step": 1000
|
267 |
+
},
|
268 |
+
{
|
269 |
+
"epoch": 0.34,
|
270 |
+
"learning_rate": 7.916e-06,
|
271 |
+
"loss": 0.1148,
|
272 |
+
"step": 1025
|
273 |
+
},
|
274 |
+
{
|
275 |
+
"epoch": 0.35,
|
276 |
+
"learning_rate": 7.816000000000001e-06,
|
277 |
+
"loss": 0.111,
|
278 |
+
"step": 1050
|
279 |
+
},
|
280 |
+
{
|
281 |
+
"epoch": 0.36,
|
282 |
+
"learning_rate": 7.716e-06,
|
283 |
+
"loss": 0.1198,
|
284 |
+
"step": 1075
|
285 |
+
},
|
286 |
+
{
|
287 |
+
"epoch": 0.37,
|
288 |
+
"learning_rate": 7.616000000000001e-06,
|
289 |
+
"loss": 0.1192,
|
290 |
+
"step": 1100
|
291 |
+
},
|
292 |
+
{
|
293 |
+
"epoch": 0.38,
|
294 |
+
"learning_rate": 7.516000000000001e-06,
|
295 |
+
"loss": 0.1093,
|
296 |
+
"step": 1125
|
297 |
+
},
|
298 |
+
{
|
299 |
+
"epoch": 0.38,
|
300 |
+
"learning_rate": 7.416000000000001e-06,
|
301 |
+
"loss": 0.1012,
|
302 |
+
"step": 1150
|
303 |
+
},
|
304 |
+
{
|
305 |
+
"epoch": 0.39,
|
306 |
+
"learning_rate": 7.316000000000001e-06,
|
307 |
+
"loss": 0.1219,
|
308 |
+
"step": 1175
|
309 |
+
},
|
310 |
+
{
|
311 |
+
"epoch": 0.4,
|
312 |
+
"learning_rate": 7.216000000000001e-06,
|
313 |
+
"loss": 0.108,
|
314 |
+
"step": 1200
|
315 |
+
},
|
316 |
+
{
|
317 |
+
"epoch": 0.41,
|
318 |
+
"learning_rate": 7.116000000000001e-06,
|
319 |
+
"loss": 0.117,
|
320 |
+
"step": 1225
|
321 |
+
},
|
322 |
+
{
|
323 |
+
"epoch": 0.42,
|
324 |
+
"learning_rate": 7.016e-06,
|
325 |
+
"loss": 0.1139,
|
326 |
+
"step": 1250
|
327 |
+
},
|
328 |
+
{
|
329 |
+
"epoch": 0.42,
|
330 |
+
"learning_rate": 6.916e-06,
|
331 |
+
"loss": 0.1117,
|
332 |
+
"step": 1275
|
333 |
+
},
|
334 |
+
{
|
335 |
+
"epoch": 0.43,
|
336 |
+
"learning_rate": 6.8160000000000005e-06,
|
337 |
+
"loss": 0.1066,
|
338 |
+
"step": 1300
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"epoch": 0.44,
|
342 |
+
"learning_rate": 6.716000000000001e-06,
|
343 |
+
"loss": 0.1188,
|
344 |
+
"step": 1325
|
345 |
+
},
|
346 |
+
{
|
347 |
+
"epoch": 0.45,
|
348 |
+
"learning_rate": 6.616e-06,
|
349 |
+
"loss": 0.1115,
|
350 |
+
"step": 1350
|
351 |
+
},
|
352 |
+
{
|
353 |
+
"epoch": 0.46,
|
354 |
+
"learning_rate": 6.516e-06,
|
355 |
+
"loss": 0.0991,
|
356 |
+
"step": 1375
|
357 |
+
},
|
358 |
+
{
|
359 |
+
"epoch": 0.47,
|
360 |
+
"learning_rate": 6.416e-06,
|
361 |
+
"loss": 0.1097,
|
362 |
+
"step": 1400
|
363 |
+
},
|
364 |
+
{
|
365 |
+
"epoch": 0.47,
|
366 |
+
"learning_rate": 6.316000000000001e-06,
|
367 |
+
"loss": 0.1064,
|
368 |
+
"step": 1425
|
369 |
+
},
|
370 |
+
{
|
371 |
+
"epoch": 0.48,
|
372 |
+
"learning_rate": 6.216000000000001e-06,
|
373 |
+
"loss": 0.1214,
|
374 |
+
"step": 1450
|
375 |
+
},
|
376 |
+
{
|
377 |
+
"epoch": 0.49,
|
378 |
+
"learning_rate": 6.116000000000001e-06,
|
379 |
+
"loss": 0.1139,
|
380 |
+
"step": 1475
|
381 |
+
},
|
382 |
+
{
|
383 |
+
"epoch": 0.5,
|
384 |
+
"learning_rate": 6.0160000000000005e-06,
|
385 |
+
"loss": 0.1132,
|
386 |
+
"step": 1500
|
387 |
+
},
|
388 |
+
{
|
389 |
+
"epoch": 0.5,
|
390 |
+
"eval_loss": 0.1435445249080658,
|
391 |
+
"eval_runtime": 17236.1573,
|
392 |
+
"eval_samples_per_second": 0.933,
|
393 |
+
"eval_steps_per_second": 0.117,
|
394 |
+
"eval_wer": 6.882050867332498,
|
395 |
+
"step": 1500
|
396 |
+
},
|
397 |
+
{
|
398 |
+
"epoch": 0.51,
|
399 |
+
"learning_rate": 5.916000000000001e-06,
|
400 |
+
"loss": 0.0995,
|
401 |
+
"step": 1525
|
402 |
+
},
|
403 |
+
{
|
404 |
+
"epoch": 0.52,
|
405 |
+
"learning_rate": 5.816000000000001e-06,
|
406 |
+
"loss": 0.1168,
|
407 |
+
"step": 1550
|
408 |
+
},
|
409 |
+
{
|
410 |
+
"epoch": 0.53,
|
411 |
+
"learning_rate": 5.716000000000001e-06,
|
412 |
+
"loss": 0.102,
|
413 |
+
"step": 1575
|
414 |
+
},
|
415 |
+
{
|
416 |
+
"epoch": 0.53,
|
417 |
+
"learning_rate": 5.616e-06,
|
418 |
+
"loss": 0.0999,
|
419 |
+
"step": 1600
|
420 |
+
},
|
421 |
+
{
|
422 |
+
"epoch": 0.54,
|
423 |
+
"learning_rate": 5.516e-06,
|
424 |
+
"loss": 0.1065,
|
425 |
+
"step": 1625
|
426 |
+
},
|
427 |
+
{
|
428 |
+
"epoch": 0.55,
|
429 |
+
"learning_rate": 5.416e-06,
|
430 |
+
"loss": 0.1092,
|
431 |
+
"step": 1650
|
432 |
+
},
|
433 |
+
{
|
434 |
+
"epoch": 0.56,
|
435 |
+
"learning_rate": 5.3160000000000004e-06,
|
436 |
+
"loss": 0.1014,
|
437 |
+
"step": 1675
|
438 |
+
},
|
439 |
+
{
|
440 |
+
"epoch": 0.57,
|
441 |
+
"learning_rate": 5.216e-06,
|
442 |
+
"loss": 0.1016,
|
443 |
+
"step": 1700
|
444 |
+
},
|
445 |
+
{
|
446 |
+
"epoch": 0.57,
|
447 |
+
"learning_rate": 5.116000000000001e-06,
|
448 |
+
"loss": 0.096,
|
449 |
+
"step": 1725
|
450 |
+
},
|
451 |
+
{
|
452 |
+
"epoch": 0.58,
|
453 |
+
"learning_rate": 5.016000000000001e-06,
|
454 |
+
"loss": 0.0948,
|
455 |
+
"step": 1750
|
456 |
+
},
|
457 |
+
{
|
458 |
+
"epoch": 0.59,
|
459 |
+
"learning_rate": 4.916e-06,
|
460 |
+
"loss": 0.1017,
|
461 |
+
"step": 1775
|
462 |
+
},
|
463 |
+
{
|
464 |
+
"epoch": 0.6,
|
465 |
+
"learning_rate": 4.816e-06,
|
466 |
+
"loss": 0.0969,
|
467 |
+
"step": 1800
|
468 |
+
},
|
469 |
+
{
|
470 |
+
"epoch": 0.61,
|
471 |
+
"learning_rate": 4.716e-06,
|
472 |
+
"loss": 0.1081,
|
473 |
+
"step": 1825
|
474 |
+
},
|
475 |
+
{
|
476 |
+
"epoch": 0.62,
|
477 |
+
"learning_rate": 4.616e-06,
|
478 |
+
"loss": 0.1128,
|
479 |
+
"step": 1850
|
480 |
+
},
|
481 |
+
{
|
482 |
+
"epoch": 0.62,
|
483 |
+
"learning_rate": 4.5160000000000005e-06,
|
484 |
+
"loss": 0.1017,
|
485 |
+
"step": 1875
|
486 |
+
},
|
487 |
+
{
|
488 |
+
"epoch": 0.63,
|
489 |
+
"learning_rate": 4.416000000000001e-06,
|
490 |
+
"loss": 0.0963,
|
491 |
+
"step": 1900
|
492 |
+
},
|
493 |
+
{
|
494 |
+
"epoch": 0.64,
|
495 |
+
"learning_rate": 4.316e-06,
|
496 |
+
"loss": 0.1004,
|
497 |
+
"step": 1925
|
498 |
+
},
|
499 |
+
{
|
500 |
+
"epoch": 0.65,
|
501 |
+
"learning_rate": 4.216e-06,
|
502 |
+
"loss": 0.1001,
|
503 |
+
"step": 1950
|
504 |
+
},
|
505 |
+
{
|
506 |
+
"epoch": 0.66,
|
507 |
+
"learning_rate": 4.116000000000001e-06,
|
508 |
+
"loss": 0.0999,
|
509 |
+
"step": 1975
|
510 |
+
},
|
511 |
+
{
|
512 |
+
"epoch": 0.67,
|
513 |
+
"learning_rate": 4.016e-06,
|
514 |
+
"loss": 0.0939,
|
515 |
+
"step": 2000
|
516 |
+
},
|
517 |
+
{
|
518 |
+
"epoch": 0.67,
|
519 |
+
"eval_loss": 0.13536924123764038,
|
520 |
+
"eval_runtime": 17221.386,
|
521 |
+
"eval_samples_per_second": 0.934,
|
522 |
+
"eval_steps_per_second": 0.117,
|
523 |
+
"eval_wer": 6.534309166632971,
|
524 |
+
"step": 2000
|
525 |
+
},
|
526 |
+
{
|
527 |
+
"epoch": 0.68,
|
528 |
+
"learning_rate": 3.916e-06,
|
529 |
+
"loss": 0.0911,
|
530 |
+
"step": 2025
|
531 |
+
},
|
532 |
+
{
|
533 |
+
"epoch": 0.68,
|
534 |
+
"learning_rate": 3.816e-06,
|
535 |
+
"loss": 0.0985,
|
536 |
+
"step": 2050
|
537 |
+
},
|
538 |
+
{
|
539 |
+
"epoch": 0.69,
|
540 |
+
"learning_rate": 3.716e-06,
|
541 |
+
"loss": 0.0954,
|
542 |
+
"step": 2075
|
543 |
+
},
|
544 |
+
{
|
545 |
+
"epoch": 0.7,
|
546 |
+
"learning_rate": 3.616e-06,
|
547 |
+
"loss": 0.0866,
|
548 |
+
"step": 2100
|
549 |
+
},
|
550 |
+
{
|
551 |
+
"epoch": 0.71,
|
552 |
+
"learning_rate": 3.5160000000000007e-06,
|
553 |
+
"loss": 0.0945,
|
554 |
+
"step": 2125
|
555 |
+
},
|
556 |
+
{
|
557 |
+
"epoch": 0.72,
|
558 |
+
"learning_rate": 3.4160000000000004e-06,
|
559 |
+
"loss": 0.1051,
|
560 |
+
"step": 2150
|
561 |
+
},
|
562 |
+
{
|
563 |
+
"epoch": 0.72,
|
564 |
+
"learning_rate": 3.3160000000000005e-06,
|
565 |
+
"loss": 0.0984,
|
566 |
+
"step": 2175
|
567 |
+
},
|
568 |
+
{
|
569 |
+
"epoch": 0.73,
|
570 |
+
"learning_rate": 3.216e-06,
|
571 |
+
"loss": 0.1037,
|
572 |
+
"step": 2200
|
573 |
+
},
|
574 |
+
{
|
575 |
+
"epoch": 0.74,
|
576 |
+
"learning_rate": 3.1160000000000003e-06,
|
577 |
+
"loss": 0.0993,
|
578 |
+
"step": 2225
|
579 |
+
},
|
580 |
+
{
|
581 |
+
"epoch": 0.75,
|
582 |
+
"learning_rate": 3.016e-06,
|
583 |
+
"loss": 0.0924,
|
584 |
+
"step": 2250
|
585 |
+
},
|
586 |
+
{
|
587 |
+
"epoch": 0.76,
|
588 |
+
"learning_rate": 2.9160000000000005e-06,
|
589 |
+
"loss": 0.0972,
|
590 |
+
"step": 2275
|
591 |
+
},
|
592 |
+
{
|
593 |
+
"epoch": 0.77,
|
594 |
+
"learning_rate": 2.8160000000000002e-06,
|
595 |
+
"loss": 0.0898,
|
596 |
+
"step": 2300
|
597 |
+
},
|
598 |
+
{
|
599 |
+
"epoch": 0.78,
|
600 |
+
"learning_rate": 2.7160000000000003e-06,
|
601 |
+
"loss": 0.0928,
|
602 |
+
"step": 2325
|
603 |
+
},
|
604 |
+
{
|
605 |
+
"epoch": 0.78,
|
606 |
+
"learning_rate": 2.616e-06,
|
607 |
+
"loss": 0.095,
|
608 |
+
"step": 2350
|
609 |
+
},
|
610 |
+
{
|
611 |
+
"epoch": 0.79,
|
612 |
+
"learning_rate": 2.516e-06,
|
613 |
+
"loss": 0.0831,
|
614 |
+
"step": 2375
|
615 |
+
},
|
616 |
+
{
|
617 |
+
"epoch": 0.8,
|
618 |
+
"learning_rate": 2.4160000000000002e-06,
|
619 |
+
"loss": 0.0853,
|
620 |
+
"step": 2400
|
621 |
+
},
|
622 |
+
{
|
623 |
+
"epoch": 0.81,
|
624 |
+
"learning_rate": 2.3160000000000004e-06,
|
625 |
+
"loss": 0.0845,
|
626 |
+
"step": 2425
|
627 |
+
},
|
628 |
+
{
|
629 |
+
"epoch": 0.82,
|
630 |
+
"learning_rate": 2.216e-06,
|
631 |
+
"loss": 0.08,
|
632 |
+
"step": 2450
|
633 |
+
},
|
634 |
+
{
|
635 |
+
"epoch": 0.82,
|
636 |
+
"learning_rate": 2.116e-06,
|
637 |
+
"loss": 0.0873,
|
638 |
+
"step": 2475
|
639 |
+
},
|
640 |
+
{
|
641 |
+
"epoch": 0.83,
|
642 |
+
"learning_rate": 2.02e-06,
|
643 |
+
"loss": 0.0819,
|
644 |
+
"step": 2500
|
645 |
+
},
|
646 |
+
{
|
647 |
+
"epoch": 0.83,
|
648 |
+
"eval_loss": 0.13387082517147064,
|
649 |
+
"eval_runtime": 17283.6443,
|
650 |
+
"eval_samples_per_second": 0.93,
|
651 |
+
"eval_steps_per_second": 0.116,
|
652 |
+
"eval_wer": 6.497917593303951,
|
653 |
+
"step": 2500
|
654 |
+
},
|
655 |
+
{
|
656 |
+
"epoch": 0.84,
|
657 |
+
"learning_rate": 1.9200000000000003e-06,
|
658 |
+
"loss": 0.0903,
|
659 |
+
"step": 2525
|
660 |
+
},
|
661 |
+
{
|
662 |
+
"epoch": 0.85,
|
663 |
+
"learning_rate": 1.8200000000000002e-06,
|
664 |
+
"loss": 0.0923,
|
665 |
+
"step": 2550
|
666 |
+
},
|
667 |
+
{
|
668 |
+
"epoch": 0.86,
|
669 |
+
"learning_rate": 1.72e-06,
|
670 |
+
"loss": 0.0855,
|
671 |
+
"step": 2575
|
672 |
+
},
|
673 |
+
{
|
674 |
+
"epoch": 0.87,
|
675 |
+
"learning_rate": 1.6200000000000002e-06,
|
676 |
+
"loss": 0.088,
|
677 |
+
"step": 2600
|
678 |
+
},
|
679 |
+
{
|
680 |
+
"epoch": 0.88,
|
681 |
+
"learning_rate": 1.52e-06,
|
682 |
+
"loss": 0.0833,
|
683 |
+
"step": 2625
|
684 |
+
},
|
685 |
+
{
|
686 |
+
"epoch": 0.88,
|
687 |
+
"learning_rate": 1.42e-06,
|
688 |
+
"loss": 0.0879,
|
689 |
+
"step": 2650
|
690 |
+
},
|
691 |
+
{
|
692 |
+
"epoch": 0.89,
|
693 |
+
"learning_rate": 1.32e-06,
|
694 |
+
"loss": 0.0951,
|
695 |
+
"step": 2675
|
696 |
+
},
|
697 |
+
{
|
698 |
+
"epoch": 0.9,
|
699 |
+
"learning_rate": 1.2200000000000002e-06,
|
700 |
+
"loss": 0.0945,
|
701 |
+
"step": 2700
|
702 |
+
},
|
703 |
+
{
|
704 |
+
"epoch": 0.91,
|
705 |
+
"learning_rate": 1.12e-06,
|
706 |
+
"loss": 0.0828,
|
707 |
+
"step": 2725
|
708 |
+
},
|
709 |
+
{
|
710 |
+
"epoch": 0.92,
|
711 |
+
"learning_rate": 1.02e-06,
|
712 |
+
"loss": 0.0893,
|
713 |
+
"step": 2750
|
714 |
+
},
|
715 |
+
{
|
716 |
+
"epoch": 0.93,
|
717 |
+
"learning_rate": 9.200000000000001e-07,
|
718 |
+
"loss": 0.082,
|
719 |
+
"step": 2775
|
720 |
+
},
|
721 |
+
{
|
722 |
+
"epoch": 0.93,
|
723 |
+
"learning_rate": 8.200000000000001e-07,
|
724 |
+
"loss": 0.0915,
|
725 |
+
"step": 2800
|
726 |
+
},
|
727 |
+
{
|
728 |
+
"epoch": 0.94,
|
729 |
+
"learning_rate": 7.2e-07,
|
730 |
+
"loss": 0.095,
|
731 |
+
"step": 2825
|
732 |
+
},
|
733 |
+
{
|
734 |
+
"epoch": 0.95,
|
735 |
+
"learning_rate": 6.200000000000001e-07,
|
736 |
+
"loss": 0.0921,
|
737 |
+
"step": 2850
|
738 |
+
},
|
739 |
+
{
|
740 |
+
"epoch": 0.96,
|
741 |
+
"learning_rate": 5.2e-07,
|
742 |
+
"loss": 0.0839,
|
743 |
+
"step": 2875
|
744 |
+
},
|
745 |
+
{
|
746 |
+
"epoch": 0.97,
|
747 |
+
"learning_rate": 4.2000000000000006e-07,
|
748 |
+
"loss": 0.0944,
|
749 |
+
"step": 2900
|
750 |
+
},
|
751 |
+
{
|
752 |
+
"epoch": 0.97,
|
753 |
+
"learning_rate": 3.2e-07,
|
754 |
+
"loss": 0.0871,
|
755 |
+
"step": 2925
|
756 |
+
},
|
757 |
+
{
|
758 |
+
"epoch": 0.98,
|
759 |
+
"learning_rate": 2.2e-07,
|
760 |
+
"loss": 0.0834,
|
761 |
+
"step": 2950
|
762 |
+
},
|
763 |
+
{
|
764 |
+
"epoch": 0.99,
|
765 |
+
"learning_rate": 1.2000000000000002e-07,
|
766 |
+
"loss": 0.089,
|
767 |
+
"step": 2975
|
768 |
+
},
|
769 |
+
{
|
770 |
+
"epoch": 1.0,
|
771 |
+
"learning_rate": 2e-08,
|
772 |
+
"loss": 0.0892,
|
773 |
+
"step": 3000
|
774 |
+
},
|
775 |
+
{
|
776 |
+
"epoch": 1.0,
|
777 |
+
"eval_loss": 0.13065485656261444,
|
778 |
+
"eval_runtime": 17232.2238,
|
779 |
+
"eval_samples_per_second": 0.933,
|
780 |
+
"eval_steps_per_second": 0.117,
|
781 |
+
"eval_wer": 6.313937972585015,
|
782 |
+
"step": 3000
|
783 |
+
},
|
784 |
+
{
|
785 |
+
"epoch": 1.0,
|
786 |
+
"step": 3000,
|
787 |
+
"total_flos": 4.076488359936e+20,
|
788 |
+
"train_loss": 0.12058185605208079,
|
789 |
+
"train_runtime": 159323.6173,
|
790 |
+
"train_samples_per_second": 1.205,
|
791 |
+
"train_steps_per_second": 0.019
|
792 |
+
}
|
793 |
+
],
|
794 |
+
"max_steps": 3000,
|
795 |
+
"num_train_epochs": 9223372036854775807,
|
796 |
+
"total_flos": 4.076488359936e+20,
|
797 |
+
"trial_name": null,
|
798 |
+
"trial_params": null
|
799 |
+
}
|