Upload trainer_state.json
Browse files- trainer_state.json +1096 -0
trainer_state.json
ADDED
@@ -0,0 +1,1096 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": 0.15353840099691743,
|
3 |
+
"best_model_checkpoint": "/scratch/lingjzhu_root/lingjzhu1/lingjzhu/g2p/byt5_small_finetuned_from_pretrained/checkpoint-5000",
|
4 |
+
"epoch": 9.646302250803858,
|
5 |
+
"global_step": 135000,
|
6 |
+
"is_hyper_param_search": false,
|
7 |
+
"is_local_process_zero": true,
|
8 |
+
"is_world_process_zero": true,
|
9 |
+
"log_history": [
|
10 |
+
{
|
11 |
+
"epoch": 0.07,
|
12 |
+
"learning_rate": 0.0003,
|
13 |
+
"loss": 1.3832,
|
14 |
+
"step": 1000
|
15 |
+
},
|
16 |
+
{
|
17 |
+
"epoch": 0.14,
|
18 |
+
"learning_rate": 0.0002999616623572683,
|
19 |
+
"loss": 0.2586,
|
20 |
+
"step": 2000
|
21 |
+
},
|
22 |
+
{
|
23 |
+
"epoch": 0.21,
|
24 |
+
"learning_rate": 0.00029984666902607135,
|
25 |
+
"loss": 0.1824,
|
26 |
+
"step": 3000
|
27 |
+
},
|
28 |
+
{
|
29 |
+
"epoch": 0.29,
|
30 |
+
"learning_rate": 0.0002996550787873857,
|
31 |
+
"loss": 0.152,
|
32 |
+
"step": 4000
|
33 |
+
},
|
34 |
+
{
|
35 |
+
"epoch": 0.36,
|
36 |
+
"learning_rate": 0.0002993869895761197,
|
37 |
+
"loss": 0.1328,
|
38 |
+
"step": 5000
|
39 |
+
},
|
40 |
+
{
|
41 |
+
"epoch": 0.36,
|
42 |
+
"eval_cer": 0.15353840099691743,
|
43 |
+
"eval_loss": 0.16419939696788788,
|
44 |
+
"eval_runtime": 90.9116,
|
45 |
+
"eval_samples_per_second": 54.449,
|
46 |
+
"eval_steps_per_second": 0.429,
|
47 |
+
"eval_wer": 0.40606060606060607,
|
48 |
+
"step": 5000
|
49 |
+
},
|
50 |
+
{
|
51 |
+
"epoch": 0.43,
|
52 |
+
"learning_rate": 0.000299042538431052,
|
53 |
+
"loss": 0.1171,
|
54 |
+
"step": 6000
|
55 |
+
},
|
56 |
+
{
|
57 |
+
"epoch": 0.5,
|
58 |
+
"learning_rate": 0.00029862190142478177,
|
59 |
+
"loss": 0.1067,
|
60 |
+
"step": 7000
|
61 |
+
},
|
62 |
+
{
|
63 |
+
"epoch": 0.57,
|
64 |
+
"learning_rate": 0.00029812529357372587,
|
65 |
+
"loss": 0.0979,
|
66 |
+
"step": 8000
|
67 |
+
},
|
68 |
+
{
|
69 |
+
"epoch": 0.64,
|
70 |
+
"learning_rate": 0.00029755296872820933,
|
71 |
+
"loss": 0.0911,
|
72 |
+
"step": 9000
|
73 |
+
},
|
74 |
+
{
|
75 |
+
"epoch": 0.71,
|
76 |
+
"learning_rate": 0.0002969052194427048,
|
77 |
+
"loss": 0.0857,
|
78 |
+
"step": 10000
|
79 |
+
},
|
80 |
+
{
|
81 |
+
"epoch": 0.71,
|
82 |
+
"eval_cer": 0.1219912113858464,
|
83 |
+
"eval_loss": 0.10491432994604111,
|
84 |
+
"eval_runtime": 90.8023,
|
85 |
+
"eval_samples_per_second": 54.514,
|
86 |
+
"eval_steps_per_second": 0.43,
|
87 |
+
"eval_wer": 0.3408080808080808,
|
88 |
+
"step": 10000
|
89 |
+
},
|
90 |
+
{
|
91 |
+
"epoch": 0.79,
|
92 |
+
"learning_rate": 0.0002961823768262882,
|
93 |
+
"loss": 0.082,
|
94 |
+
"step": 11000
|
95 |
+
},
|
96 |
+
{
|
97 |
+
"epoch": 0.86,
|
98 |
+
"learning_rate": 0.0002953848103733858,
|
99 |
+
"loss": 0.0774,
|
100 |
+
"step": 12000
|
101 |
+
},
|
102 |
+
{
|
103 |
+
"epoch": 0.93,
|
104 |
+
"learning_rate": 0.00029451292777490066,
|
105 |
+
"loss": 0.0745,
|
106 |
+
"step": 13000
|
107 |
+
},
|
108 |
+
{
|
109 |
+
"epoch": 1.0,
|
110 |
+
"learning_rate": 0.0002935671747098137,
|
111 |
+
"loss": 0.0708,
|
112 |
+
"step": 14000
|
113 |
+
},
|
114 |
+
{
|
115 |
+
"epoch": 1.07,
|
116 |
+
"learning_rate": 0.00029254803461736643,
|
117 |
+
"loss": 0.0673,
|
118 |
+
"step": 15000
|
119 |
+
},
|
120 |
+
{
|
121 |
+
"epoch": 1.07,
|
122 |
+
"eval_cer": 0.11350866837192015,
|
123 |
+
"eval_loss": 0.08455779403448105,
|
124 |
+
"eval_runtime": 91.1047,
|
125 |
+
"eval_samples_per_second": 54.333,
|
126 |
+
"eval_steps_per_second": 0.428,
|
127 |
+
"eval_wer": 0.31757575757575757,
|
128 |
+
"step": 15000
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 1.14,
|
132 |
+
"learning_rate": 0.00029145602844994243,
|
133 |
+
"loss": 0.065,
|
134 |
+
"step": 16000
|
135 |
+
},
|
136 |
+
{
|
137 |
+
"epoch": 1.21,
|
138 |
+
"learning_rate": 0.0002902917144067724,
|
139 |
+
"loss": 0.0625,
|
140 |
+
"step": 17000
|
141 |
+
},
|
142 |
+
{
|
143 |
+
"epoch": 1.29,
|
144 |
+
"learning_rate": 0.00028905568764860047,
|
145 |
+
"loss": 0.0611,
|
146 |
+
"step": 18000
|
147 |
+
},
|
148 |
+
{
|
149 |
+
"epoch": 1.36,
|
150 |
+
"learning_rate": 0.00028774857999345685,
|
151 |
+
"loss": 0.0599,
|
152 |
+
"step": 19000
|
153 |
+
},
|
154 |
+
{
|
155 |
+
"epoch": 1.43,
|
156 |
+
"learning_rate": 0.0002863710595936922,
|
157 |
+
"loss": 0.0585,
|
158 |
+
"step": 20000
|
159 |
+
},
|
160 |
+
{
|
161 |
+
"epoch": 1.43,
|
162 |
+
"eval_cer": 0.10635972103801841,
|
163 |
+
"eval_loss": 0.0741606280207634,
|
164 |
+
"eval_runtime": 90.9345,
|
165 |
+
"eval_samples_per_second": 54.435,
|
166 |
+
"eval_steps_per_second": 0.429,
|
167 |
+
"eval_wer": 0.29757575757575755,
|
168 |
+
"step": 20000
|
169 |
+
},
|
170 |
+
{
|
171 |
+
"epoch": 1.5,
|
172 |
+
"learning_rate": 0.0002849238305944389,
|
173 |
+
"loss": 0.057,
|
174 |
+
"step": 21000
|
175 |
+
},
|
176 |
+
{
|
177 |
+
"epoch": 1.57,
|
178 |
+
"learning_rate": 0.00028340763277367477,
|
179 |
+
"loss": 0.0557,
|
180 |
+
"step": 22000
|
181 |
+
},
|
182 |
+
{
|
183 |
+
"epoch": 1.64,
|
184 |
+
"learning_rate": 0.0002818232411640713,
|
185 |
+
"loss": 0.0548,
|
186 |
+
"step": 23000
|
187 |
+
},
|
188 |
+
{
|
189 |
+
"epoch": 1.71,
|
190 |
+
"learning_rate": 0.00028017146565682144,
|
191 |
+
"loss": 0.0538,
|
192 |
+
"step": 24000
|
193 |
+
},
|
194 |
+
{
|
195 |
+
"epoch": 1.79,
|
196 |
+
"learning_rate": 0.00027845315058764886,
|
197 |
+
"loss": 0.0526,
|
198 |
+
"step": 25000
|
199 |
+
},
|
200 |
+
{
|
201 |
+
"epoch": 1.79,
|
202 |
+
"eval_cer": 0.10283990293172428,
|
203 |
+
"eval_loss": 0.06704817712306976,
|
204 |
+
"eval_runtime": 91.2941,
|
205 |
+
"eval_samples_per_second": 54.22,
|
206 |
+
"eval_steps_per_second": 0.427,
|
207 |
+
"eval_wer": 0.2898989898989899,
|
208 |
+
"step": 25000
|
209 |
+
},
|
210 |
+
{
|
211 |
+
"epoch": 1.86,
|
212 |
+
"learning_rate": 0.00027666917430520975,
|
213 |
+
"loss": 0.0519,
|
214 |
+
"step": 26000
|
215 |
+
},
|
216 |
+
{
|
217 |
+
"epoch": 1.93,
|
218 |
+
"learning_rate": 0.00027482044872210895,
|
219 |
+
"loss": 0.0506,
|
220 |
+
"step": 27000
|
221 |
+
},
|
222 |
+
{
|
223 |
+
"epoch": 2.0,
|
224 |
+
"learning_rate": 0.0002729079188487587,
|
225 |
+
"loss": 0.0498,
|
226 |
+
"step": 28000
|
227 |
+
},
|
228 |
+
{
|
229 |
+
"epoch": 2.07,
|
230 |
+
"learning_rate": 0.00027093256231031885,
|
231 |
+
"loss": 0.0468,
|
232 |
+
"step": 29000
|
233 |
+
},
|
234 |
+
{
|
235 |
+
"epoch": 2.14,
|
236 |
+
"learning_rate": 0.00026889538884696597,
|
237 |
+
"loss": 0.0463,
|
238 |
+
"step": 30000
|
239 |
+
},
|
240 |
+
{
|
241 |
+
"epoch": 2.14,
|
242 |
+
"eval_cer": 0.10310224962287663,
|
243 |
+
"eval_loss": 0.0634964108467102,
|
244 |
+
"eval_runtime": 91.3121,
|
245 |
+
"eval_samples_per_second": 54.21,
|
246 |
+
"eval_steps_per_second": 0.427,
|
247 |
+
"eval_wer": 0.2896969696969697,
|
248 |
+
"step": 30000
|
249 |
+
},
|
250 |
+
{
|
251 |
+
"epoch": 2.22,
|
252 |
+
"learning_rate": 0.0002667974397977457,
|
253 |
+
"loss": 0.0461,
|
254 |
+
"step": 31000
|
255 |
+
},
|
256 |
+
{
|
257 |
+
"epoch": 2.29,
|
258 |
+
"learning_rate": 0.0002646397875682729,
|
259 |
+
"loss": 0.0454,
|
260 |
+
"step": 32000
|
261 |
+
},
|
262 |
+
{
|
263 |
+
"epoch": 2.36,
|
264 |
+
"learning_rate": 0.00026242353508255185,
|
265 |
+
"loss": 0.0448,
|
266 |
+
"step": 33000
|
267 |
+
},
|
268 |
+
{
|
269 |
+
"epoch": 2.43,
|
270 |
+
"learning_rate": 0.0002601498152191957,
|
271 |
+
"loss": 0.0446,
|
272 |
+
"step": 34000
|
273 |
+
},
|
274 |
+
{
|
275 |
+
"epoch": 2.5,
|
276 |
+
"learning_rate": 0.0002578197902323352,
|
277 |
+
"loss": 0.0443,
|
278 |
+
"step": 35000
|
279 |
+
},
|
280 |
+
{
|
281 |
+
"epoch": 2.5,
|
282 |
+
"eval_cer": 0.1002382982444634,
|
283 |
+
"eval_loss": 0.06088022515177727,
|
284 |
+
"eval_runtime": 91.0524,
|
285 |
+
"eval_samples_per_second": 54.364,
|
286 |
+
"eval_steps_per_second": 0.428,
|
287 |
+
"eval_wer": 0.282020202020202,
|
288 |
+
"step": 35000
|
289 |
+
},
|
290 |
+
{
|
291 |
+
"epoch": 2.57,
|
292 |
+
"learning_rate": 0.00025543465115751026,
|
293 |
+
"loss": 0.0434,
|
294 |
+
"step": 36000
|
295 |
+
},
|
296 |
+
{
|
297 |
+
"epoch": 2.64,
|
298 |
+
"learning_rate": 0.0002529956172028505,
|
299 |
+
"loss": 0.0431,
|
300 |
+
"step": 37000
|
301 |
+
},
|
302 |
+
{
|
303 |
+
"epoch": 2.72,
|
304 |
+
"learning_rate": 0.0002505039351258541,
|
305 |
+
"loss": 0.0426,
|
306 |
+
"step": 38000
|
307 |
+
},
|
308 |
+
{
|
309 |
+
"epoch": 2.79,
|
310 |
+
"learning_rate": 0.0002479608785960846,
|
311 |
+
"loss": 0.0417,
|
312 |
+
"step": 39000
|
313 |
+
},
|
314 |
+
{
|
315 |
+
"epoch": 2.86,
|
316 |
+
"learning_rate": 0.0002453677475441111,
|
317 |
+
"loss": 0.0415,
|
318 |
+
"step": 40000
|
319 |
+
},
|
320 |
+
{
|
321 |
+
"epoch": 2.86,
|
322 |
+
"eval_cer": 0.09859863142476115,
|
323 |
+
"eval_loss": 0.05760909989476204,
|
324 |
+
"eval_runtime": 91.2335,
|
325 |
+
"eval_samples_per_second": 54.256,
|
326 |
+
"eval_steps_per_second": 0.427,
|
327 |
+
"eval_wer": 0.27494949494949494,
|
328 |
+
"step": 40000
|
329 |
+
},
|
330 |
+
{
|
331 |
+
"epoch": 2.93,
|
332 |
+
"learning_rate": 0.00024272586749702474,
|
333 |
+
"loss": 0.0411,
|
334 |
+
"step": 41000
|
335 |
+
},
|
336 |
+
{
|
337 |
+
"epoch": 3.0,
|
338 |
+
"learning_rate": 0.0002400365889008706,
|
339 |
+
"loss": 0.0408,
|
340 |
+
"step": 42000
|
341 |
+
},
|
342 |
+
{
|
343 |
+
"epoch": 3.07,
|
344 |
+
"learning_rate": 0.00023730128643034235,
|
345 |
+
"loss": 0.0377,
|
346 |
+
"step": 43000
|
347 |
+
},
|
348 |
+
{
|
349 |
+
"epoch": 3.14,
|
350 |
+
"learning_rate": 0.00023452135828609167,
|
351 |
+
"loss": 0.0381,
|
352 |
+
"step": 44000
|
353 |
+
},
|
354 |
+
{
|
355 |
+
"epoch": 3.22,
|
356 |
+
"learning_rate": 0.0002316982254800121,
|
357 |
+
"loss": 0.0379,
|
358 |
+
"step": 45000
|
359 |
+
},
|
360 |
+
{
|
361 |
+
"epoch": 3.22,
|
362 |
+
"eval_cer": 0.09984477820773485,
|
363 |
+
"eval_loss": 0.056990817189216614,
|
364 |
+
"eval_runtime": 91.2394,
|
365 |
+
"eval_samples_per_second": 54.253,
|
366 |
+
"eval_steps_per_second": 0.427,
|
367 |
+
"eval_wer": 0.27595959595959596,
|
368 |
+
"step": 45000
|
369 |
+
},
|
370 |
+
{
|
371 |
+
"epoch": 3.29,
|
372 |
+
"learning_rate": 0.00022883333110886237,
|
373 |
+
"loss": 0.0375,
|
374 |
+
"step": 46000
|
375 |
+
},
|
376 |
+
{
|
377 |
+
"epoch": 3.36,
|
378 |
+
"learning_rate": 0.00022592813961660067,
|
379 |
+
"loss": 0.0372,
|
380 |
+
"step": 47000
|
381 |
+
},
|
382 |
+
{
|
383 |
+
"epoch": 3.43,
|
384 |
+
"learning_rate": 0.00022298413604580696,
|
385 |
+
"loss": 0.037,
|
386 |
+
"step": 48000
|
387 |
+
},
|
388 |
+
{
|
389 |
+
"epoch": 3.5,
|
390 |
+
"learning_rate": 0.00022000282527857588,
|
391 |
+
"loss": 0.0367,
|
392 |
+
"step": 49000
|
393 |
+
},
|
394 |
+
{
|
395 |
+
"epoch": 3.57,
|
396 |
+
"learning_rate": 0.0002169857312672683,
|
397 |
+
"loss": 0.0366,
|
398 |
+
"step": 50000
|
399 |
+
},
|
400 |
+
{
|
401 |
+
"epoch": 3.57,
|
402 |
+
"eval_cer": 0.09800835136966835,
|
403 |
+
"eval_loss": 0.05443936586380005,
|
404 |
+
"eval_runtime": 91.3792,
|
405 |
+
"eval_samples_per_second": 54.17,
|
406 |
+
"eval_steps_per_second": 0.427,
|
407 |
+
"eval_wer": 0.27434343434343433,
|
408 |
+
"step": 50000
|
409 |
+
},
|
410 |
+
{
|
411 |
+
"epoch": 3.64,
|
412 |
+
"learning_rate": 0.00021393439625551483,
|
413 |
+
"loss": 0.0365,
|
414 |
+
"step": 51000
|
415 |
+
},
|
416 |
+
{
|
417 |
+
"epoch": 3.72,
|
418 |
+
"learning_rate": 0.00021085037998986924,
|
419 |
+
"loss": 0.036,
|
420 |
+
"step": 52000
|
421 |
+
},
|
422 |
+
{
|
423 |
+
"epoch": 3.79,
|
424 |
+
"learning_rate": 0.00020773525892251514,
|
425 |
+
"loss": 0.0361,
|
426 |
+
"step": 53000
|
427 |
+
},
|
428 |
+
{
|
429 |
+
"epoch": 3.86,
|
430 |
+
"learning_rate": 0.00020459062540543316,
|
431 |
+
"loss": 0.0358,
|
432 |
+
"step": 54000
|
433 |
+
},
|
434 |
+
{
|
435 |
+
"epoch": 3.93,
|
436 |
+
"learning_rate": 0.00020141808687644067,
|
437 |
+
"loss": 0.0355,
|
438 |
+
"step": 55000
|
439 |
+
},
|
440 |
+
{
|
441 |
+
"epoch": 3.93,
|
442 |
+
"eval_cer": 0.09678406681095734,
|
443 |
+
"eval_loss": 0.052644312381744385,
|
444 |
+
"eval_runtime": 91.3207,
|
445 |
+
"eval_samples_per_second": 54.205,
|
446 |
+
"eval_steps_per_second": 0.427,
|
447 |
+
"eval_wer": 0.27090909090909093,
|
448 |
+
"step": 55000
|
449 |
+
},
|
450 |
+
{
|
451 |
+
"epoch": 4.0,
|
452 |
+
"learning_rate": 0.00019821926503751995,
|
453 |
+
"loss": 0.0351,
|
454 |
+
"step": 56000
|
455 |
+
},
|
456 |
+
{
|
457 |
+
"epoch": 4.07,
|
458 |
+
"learning_rate": 0.00019499579502585537,
|
459 |
+
"loss": 0.0327,
|
460 |
+
"step": 57000
|
461 |
+
},
|
462 |
+
{
|
463 |
+
"epoch": 4.14,
|
464 |
+
"learning_rate": 0.00019174932457800242,
|
465 |
+
"loss": 0.0326,
|
466 |
+
"step": 58000
|
467 |
+
},
|
468 |
+
{
|
469 |
+
"epoch": 4.22,
|
470 |
+
"learning_rate": 0.0001884815131876167,
|
471 |
+
"loss": 0.0327,
|
472 |
+
"step": 59000
|
473 |
+
},
|
474 |
+
{
|
475 |
+
"epoch": 4.29,
|
476 |
+
"learning_rate": 0.00018519403125717278,
|
477 |
+
"loss": 0.0324,
|
478 |
+
"step": 60000
|
479 |
+
},
|
480 |
+
{
|
481 |
+
"epoch": 4.29,
|
482 |
+
"eval_cer": 0.09630309787717803,
|
483 |
+
"eval_loss": 0.05228239297866821,
|
484 |
+
"eval_runtime": 91.3618,
|
485 |
+
"eval_samples_per_second": 54.18,
|
486 |
+
"eval_steps_per_second": 0.427,
|
487 |
+
"eval_wer": 0.26686868686868687,
|
488 |
+
"step": 60000
|
489 |
+
},
|
490 |
+
{
|
491 |
+
"epoch": 4.36,
|
492 |
+
"learning_rate": 0.00018188855924410722,
|
493 |
+
"loss": 0.0326,
|
494 |
+
"step": 61000
|
495 |
+
},
|
496 |
+
{
|
497 |
+
"epoch": 4.43,
|
498 |
+
"learning_rate": 0.00017856678680182127,
|
499 |
+
"loss": 0.0323,
|
500 |
+
"step": 62000
|
501 |
+
},
|
502 |
+
{
|
503 |
+
"epoch": 4.5,
|
504 |
+
"learning_rate": 0.0001752304119159834,
|
505 |
+
"loss": 0.0322,
|
506 |
+
"step": 63000
|
507 |
+
},
|
508 |
+
{
|
509 |
+
"epoch": 4.57,
|
510 |
+
"learning_rate": 0.00017188114003657205,
|
511 |
+
"loss": 0.0323,
|
512 |
+
"step": 64000
|
513 |
+
},
|
514 |
+
{
|
515 |
+
"epoch": 4.64,
|
516 |
+
"learning_rate": 0.00016852068320610358,
|
517 |
+
"loss": 0.0316,
|
518 |
+
"step": 65000
|
519 |
+
},
|
520 |
+
{
|
521 |
+
"epoch": 4.64,
|
522 |
+
"eval_cer": 0.09446667103911152,
|
523 |
+
"eval_loss": 0.050846464931964874,
|
524 |
+
"eval_runtime": 91.2629,
|
525 |
+
"eval_samples_per_second": 54.239,
|
526 |
+
"eval_steps_per_second": 0.427,
|
527 |
+
"eval_wer": 0.2638383838383838,
|
528 |
+
"step": 65000
|
529 |
+
},
|
530 |
+
{
|
531 |
+
"epoch": 4.72,
|
532 |
+
"learning_rate": 0.00016515075918448972,
|
533 |
+
"loss": 0.0318,
|
534 |
+
"step": 66000
|
535 |
+
},
|
536 |
+
{
|
537 |
+
"epoch": 4.79,
|
538 |
+
"learning_rate": 0.00016177309057097285,
|
539 |
+
"loss": 0.0315,
|
540 |
+
"step": 67000
|
541 |
+
},
|
542 |
+
{
|
543 |
+
"epoch": 4.86,
|
544 |
+
"learning_rate": 0.00015838940392358722,
|
545 |
+
"loss": 0.0315,
|
546 |
+
"step": 68000
|
547 |
+
},
|
548 |
+
{
|
549 |
+
"epoch": 4.93,
|
550 |
+
"learning_rate": 0.00015500142887659688,
|
551 |
+
"loss": 0.0313,
|
552 |
+
"step": 69000
|
553 |
+
},
|
554 |
+
{
|
555 |
+
"epoch": 5.0,
|
556 |
+
"learning_rate": 0.00015161089725636095,
|
557 |
+
"loss": 0.0313,
|
558 |
+
"step": 70000
|
559 |
+
},
|
560 |
+
{
|
561 |
+
"epoch": 5.0,
|
562 |
+
"eval_cer": 0.09420432434795917,
|
563 |
+
"eval_loss": 0.048644062131643295,
|
564 |
+
"eval_runtime": 91.4526,
|
565 |
+
"eval_samples_per_second": 54.126,
|
566 |
+
"eval_steps_per_second": 0.426,
|
567 |
+
"eval_wer": 0.26141414141414143,
|
568 |
+
"step": 70000
|
569 |
+
},
|
570 |
+
{
|
571 |
+
"epoch": 5.07,
|
572 |
+
"learning_rate": 0.00014821954219607845,
|
573 |
+
"loss": 0.0289,
|
574 |
+
"step": 71000
|
575 |
+
},
|
576 |
+
{
|
577 |
+
"epoch": 5.14,
|
578 |
+
"learning_rate": 0.0001448290972498651,
|
579 |
+
"loss": 0.029,
|
580 |
+
"step": 72000
|
581 |
+
},
|
582 |
+
{
|
583 |
+
"epoch": 5.22,
|
584 |
+
"learning_rate": 0.00014144129550661485,
|
585 |
+
"loss": 0.0288,
|
586 |
+
"step": 73000
|
587 |
+
},
|
588 |
+
{
|
589 |
+
"epoch": 5.29,
|
590 |
+
"learning_rate": 0.0001380578687040995,
|
591 |
+
"loss": 0.0289,
|
592 |
+
"step": 74000
|
593 |
+
},
|
594 |
+
{
|
595 |
+
"epoch": 5.36,
|
596 |
+
"learning_rate": 0.00013468054634375843,
|
597 |
+
"loss": 0.0292,
|
598 |
+
"step": 75000
|
599 |
+
},
|
600 |
+
{
|
601 |
+
"epoch": 5.36,
|
602 |
+
"eval_cer": 0.0939638398810695,
|
603 |
+
"eval_loss": 0.049041230231523514,
|
604 |
+
"eval_runtime": 91.4249,
|
605 |
+
"eval_samples_per_second": 54.143,
|
606 |
+
"eval_steps_per_second": 0.427,
|
607 |
+
"eval_wer": 0.258989898989899,
|
608 |
+
"step": 75000
|
609 |
+
},
|
610 |
+
{
|
611 |
+
"epoch": 5.43,
|
612 |
+
"learning_rate": 0.00013131105480663235,
|
613 |
+
"loss": 0.0289,
|
614 |
+
"step": 76000
|
615 |
+
},
|
616 |
+
{
|
617 |
+
"epoch": 5.5,
|
618 |
+
"learning_rate": 0.000127951116470891,
|
619 |
+
"loss": 0.0288,
|
620 |
+
"step": 77000
|
621 |
+
},
|
622 |
+
{
|
623 |
+
"epoch": 5.57,
|
624 |
+
"learning_rate": 0.00012460244883140783,
|
625 |
+
"loss": 0.0287,
|
626 |
+
"step": 78000
|
627 |
+
},
|
628 |
+
{
|
629 |
+
"epoch": 5.64,
|
630 |
+
"learning_rate": 0.0001212667636218309,
|
631 |
+
"loss": 0.0284,
|
632 |
+
"step": 79000
|
633 |
+
},
|
634 |
+
{
|
635 |
+
"epoch": 5.72,
|
636 |
+
"learning_rate": 0.00011794576593959775,
|
637 |
+
"loss": 0.0283,
|
638 |
+
"step": 80000
|
639 |
+
},
|
640 |
+
{
|
641 |
+
"epoch": 5.72,
|
642 |
+
"eval_cer": 0.0939419776568068,
|
643 |
+
"eval_loss": 0.04810198023915291,
|
644 |
+
"eval_runtime": 91.3828,
|
645 |
+
"eval_samples_per_second": 54.168,
|
646 |
+
"eval_steps_per_second": 0.427,
|
647 |
+
"eval_wer": 0.2604040404040404,
|
648 |
+
"step": 80000
|
649 |
+
},
|
650 |
+
{
|
651 |
+
"epoch": 5.79,
|
652 |
+
"learning_rate": 0.00011464115337434394,
|
653 |
+
"loss": 0.029,
|
654 |
+
"step": 81000
|
655 |
+
},
|
656 |
+
{
|
657 |
+
"epoch": 5.86,
|
658 |
+
"learning_rate": 0.00011135461514014796,
|
659 |
+
"loss": 0.0289,
|
660 |
+
"step": 82000
|
661 |
+
},
|
662 |
+
{
|
663 |
+
"epoch": 5.93,
|
664 |
+
"learning_rate": 0.00010808783121205837,
|
665 |
+
"loss": 0.0285,
|
666 |
+
"step": 83000
|
667 |
+
},
|
668 |
+
{
|
669 |
+
"epoch": 6.0,
|
670 |
+
"learning_rate": 0.00010484247146734352,
|
671 |
+
"loss": 0.029,
|
672 |
+
"step": 84000
|
673 |
+
},
|
674 |
+
{
|
675 |
+
"epoch": 6.07,
|
676 |
+
"learning_rate": 0.00010162019483190237,
|
677 |
+
"loss": 0.0282,
|
678 |
+
"step": 85000
|
679 |
+
},
|
680 |
+
{
|
681 |
+
"epoch": 6.07,
|
682 |
+
"eval_cer": 0.09354845762007827,
|
683 |
+
"eval_loss": 0.048511628061532974,
|
684 |
+
"eval_runtime": 93.8707,
|
685 |
+
"eval_samples_per_second": 52.732,
|
686 |
+
"eval_steps_per_second": 0.415,
|
687 |
+
"eval_wer": 0.25696969696969696,
|
688 |
+
"step": 85000
|
689 |
+
},
|
690 |
+
{
|
691 |
+
"epoch": 6.15,
|
692 |
+
"learning_rate": 9.842264843227404e-05,
|
693 |
+
"loss": 0.0282,
|
694 |
+
"step": 86000
|
695 |
+
},
|
696 |
+
{
|
697 |
+
"epoch": 6.22,
|
698 |
+
"learning_rate": 9.52514667536784e-05,
|
699 |
+
"loss": 0.0281,
|
700 |
+
"step": 87000
|
701 |
+
},
|
702 |
+
{
|
703 |
+
"epoch": 6.29,
|
704 |
+
"learning_rate": 9.210827080451842e-05,
|
705 |
+
"loss": 0.0281,
|
706 |
+
"step": 88000
|
707 |
+
},
|
708 |
+
{
|
709 |
+
"epoch": 6.36,
|
710 |
+
"learning_rate": 8.899466728777203e-05,
|
711 |
+
"loss": 0.0279,
|
712 |
+
"step": 89000
|
713 |
+
},
|
714 |
+
{
|
715 |
+
"epoch": 6.43,
|
716 |
+
"learning_rate": 8.591224777969557e-05,
|
717 |
+
"loss": 0.028,
|
718 |
+
"step": 90000
|
719 |
+
},
|
720 |
+
{
|
721 |
+
"epoch": 6.43,
|
722 |
+
"eval_cer": 0.09381080431123062,
|
723 |
+
"eval_loss": 0.047929223626852036,
|
724 |
+
"eval_runtime": 93.5944,
|
725 |
+
"eval_samples_per_second": 52.888,
|
726 |
+
"eval_steps_per_second": 0.417,
|
727 |
+
"eval_wer": 0.2608080808080808,
|
728 |
+
"step": 90000
|
729 |
+
},
|
730 |
+
{
|
731 |
+
"epoch": 6.5,
|
732 |
+
"learning_rate": 8.286258791626041e-05,
|
733 |
+
"loss": 0.028,
|
734 |
+
"step": 91000
|
735 |
+
},
|
736 |
+
{
|
737 |
+
"epoch": 6.57,
|
738 |
+
"learning_rate": 7.984724658773716e-05,
|
739 |
+
"loss": 0.0276,
|
740 |
+
"step": 92000
|
741 |
+
},
|
742 |
+
{
|
743 |
+
"epoch": 6.65,
|
744 |
+
"learning_rate": 7.686776514184009e-05,
|
745 |
+
"loss": 0.0275,
|
746 |
+
"step": 93000
|
747 |
+
},
|
748 |
+
{
|
749 |
+
"epoch": 6.72,
|
750 |
+
"learning_rate": 7.392566659583846e-05,
|
751 |
+
"loss": 0.0274,
|
752 |
+
"step": 94000
|
753 |
+
},
|
754 |
+
{
|
755 |
+
"epoch": 6.79,
|
756 |
+
"learning_rate": 7.102245485803813e-05,
|
757 |
+
"loss": 0.0276,
|
758 |
+
"step": 95000
|
759 |
+
},
|
760 |
+
{
|
761 |
+
"epoch": 6.79,
|
762 |
+
"eval_cer": 0.09346100872302748,
|
763 |
+
"eval_loss": 0.04663492366671562,
|
764 |
+
"eval_runtime": 93.7458,
|
765 |
+
"eval_samples_per_second": 52.802,
|
766 |
+
"eval_steps_per_second": 0.416,
|
767 |
+
"eval_wer": 0.25737373737373737,
|
768 |
+
"step": 95000
|
769 |
+
},
|
770 |
+
{
|
771 |
+
"epoch": 6.86,
|
772 |
+
"learning_rate": 6.81596139590308e-05,
|
773 |
+
"loss": 0.0273,
|
774 |
+
"step": 96000
|
775 |
+
},
|
776 |
+
{
|
777 |
+
"epoch": 6.93,
|
778 |
+
"learning_rate": 6.533860729310434e-05,
|
779 |
+
"loss": 0.0275,
|
780 |
+
"step": 97000
|
781 |
+
},
|
782 |
+
{
|
783 |
+
"epoch": 7.0,
|
784 |
+
"learning_rate": 6.256087687020127e-05,
|
785 |
+
"loss": 0.0271,
|
786 |
+
"step": 98000
|
787 |
+
},
|
788 |
+
{
|
789 |
+
"epoch": 7.07,
|
790 |
+
"learning_rate": 5.98278425788092e-05,
|
791 |
+
"loss": 0.0256,
|
792 |
+
"step": 99000
|
793 |
+
},
|
794 |
+
{
|
795 |
+
"epoch": 7.15,
|
796 |
+
"learning_rate": 5.71409014601578e-05,
|
797 |
+
"loss": 0.0257,
|
798 |
+
"step": 100000
|
799 |
+
},
|
800 |
+
{
|
801 |
+
"epoch": 7.15,
|
802 |
+
"eval_cer": 0.09276141754662119,
|
803 |
+
"eval_loss": 0.04688685014843941,
|
804 |
+
"eval_runtime": 93.8485,
|
805 |
+
"eval_samples_per_second": 52.745,
|
806 |
+
"eval_steps_per_second": 0.416,
|
807 |
+
"eval_wer": 0.25696969696969696,
|
808 |
+
"step": 100000
|
809 |
+
},
|
810 |
+
{
|
811 |
+
"epoch": 7.22,
|
812 |
+
"learning_rate": 5.4501426994095876e-05,
|
813 |
+
"loss": 0.0261,
|
814 |
+
"step": 101000
|
815 |
+
},
|
816 |
+
{
|
817 |
+
"epoch": 7.29,
|
818 |
+
"learning_rate": 5.191076839701103e-05,
|
819 |
+
"loss": 0.0262,
|
820 |
+
"step": 102000
|
821 |
+
},
|
822 |
+
{
|
823 |
+
"epoch": 7.36,
|
824 |
+
"learning_rate": 4.9370249932153075e-05,
|
825 |
+
"loss": 0.0261,
|
826 |
+
"step": 103000
|
827 |
+
},
|
828 |
+
{
|
829 |
+
"epoch": 7.43,
|
830 |
+
"learning_rate": 4.6881170232712164e-05,
|
831 |
+
"loss": 0.0259,
|
832 |
+
"step": 104000
|
833 |
+
},
|
834 |
+
{
|
835 |
+
"epoch": 7.5,
|
836 |
+
"learning_rate": 4.444480163799822e-05,
|
837 |
+
"loss": 0.0261,
|
838 |
+
"step": 105000
|
839 |
+
},
|
840 |
+
{
|
841 |
+
"epoch": 7.5,
|
842 |
+
"eval_cer": 0.09184320412758794,
|
843 |
+
"eval_loss": 0.04592433199286461,
|
844 |
+
"eval_runtime": 93.414,
|
845 |
+
"eval_samples_per_second": 52.99,
|
846 |
+
"eval_steps_per_second": 0.417,
|
847 |
+
"eval_wer": 0.25515151515151513,
|
848 |
+
"step": 105000
|
849 |
+
},
|
850 |
+
{
|
851 |
+
"epoch": 7.57,
|
852 |
+
"learning_rate": 4.2062389543061265e-05,
|
853 |
+
"loss": 0.0257,
|
854 |
+
"step": 106000
|
855 |
+
},
|
856 |
+
{
|
857 |
+
"epoch": 7.65,
|
858 |
+
"learning_rate": 3.9735151762084384e-05,
|
859 |
+
"loss": 0.0258,
|
860 |
+
"step": 107000
|
861 |
+
},
|
862 |
+
{
|
863 |
+
"epoch": 7.72,
|
864 |
+
"learning_rate": 3.746427790587557e-05,
|
865 |
+
"loss": 0.0258,
|
866 |
+
"step": 108000
|
867 |
+
},
|
868 |
+
{
|
869 |
+
"epoch": 7.79,
|
870 |
+
"learning_rate": 3.525092877377602e-05,
|
871 |
+
"loss": 0.0258,
|
872 |
+
"step": 109000
|
873 |
+
},
|
874 |
+
{
|
875 |
+
"epoch": 7.86,
|
876 |
+
"learning_rate": 3.309623576029597e-05,
|
877 |
+
"loss": 0.0258,
|
878 |
+
"step": 110000
|
879 |
+
},
|
880 |
+
{
|
881 |
+
"epoch": 7.86,
|
882 |
+
"eval_cer": 0.09195251524890143,
|
883 |
+
"eval_loss": 0.04556591063737869,
|
884 |
+
"eval_runtime": 93.7142,
|
885 |
+
"eval_samples_per_second": 52.82,
|
886 |
+
"eval_steps_per_second": 0.416,
|
887 |
+
"eval_wer": 0.25333333333333335,
|
888 |
+
"step": 110000
|
889 |
+
},
|
890 |
+
{
|
891 |
+
"epoch": 7.93,
|
892 |
+
"learning_rate": 3.1001300276781274e-05,
|
893 |
+
"loss": 0.0258,
|
894 |
+
"step": 111000
|
895 |
+
},
|
896 |
+
{
|
897 |
+
"epoch": 8.0,
|
898 |
+
"learning_rate": 2.8967193188406938e-05,
|
899 |
+
"loss": 0.0255,
|
900 |
+
"step": 112000
|
901 |
+
},
|
902 |
+
{
|
903 |
+
"epoch": 8.07,
|
904 |
+
"learning_rate": 2.699495426678389e-05,
|
905 |
+
"loss": 0.0241,
|
906 |
+
"step": 113000
|
907 |
+
},
|
908 |
+
{
|
909 |
+
"epoch": 8.15,
|
910 |
+
"learning_rate": 2.5085591658461056e-05,
|
911 |
+
"loss": 0.0243,
|
912 |
+
"step": 114000
|
913 |
+
},
|
914 |
+
{
|
915 |
+
"epoch": 8.22,
|
916 |
+
"learning_rate": 2.3240081369591984e-05,
|
917 |
+
"loss": 0.0249,
|
918 |
+
"step": 115000
|
919 |
+
},
|
920 |
+
{
|
921 |
+
"epoch": 8.22,
|
922 |
+
"eval_cer": 0.09116547517544435,
|
923 |
+
"eval_loss": 0.04581727460026741,
|
924 |
+
"eval_runtime": 93.453,
|
925 |
+
"eval_samples_per_second": 52.968,
|
926 |
+
"eval_steps_per_second": 0.417,
|
927 |
+
"eval_wer": 0.25292929292929295,
|
928 |
+
"step": 115000
|
929 |
+
},
|
930 |
+
{
|
931 |
+
"epoch": 8.29,
|
932 |
+
"learning_rate": 2.1459366767031522e-05,
|
933 |
+
"loss": 0.0247,
|
934 |
+
"step": 116000
|
935 |
+
},
|
936 |
+
{
|
937 |
+
"epoch": 8.36,
|
938 |
+
"learning_rate": 1.9744358096116225e-05,
|
939 |
+
"loss": 0.0249,
|
940 |
+
"step": 117000
|
941 |
+
},
|
942 |
+
{
|
943 |
+
"epoch": 8.43,
|
944 |
+
"learning_rate": 1.8095932015375496e-05,
|
945 |
+
"loss": 0.0247,
|
946 |
+
"step": 118000
|
947 |
+
},
|
948 |
+
{
|
949 |
+
"epoch": 8.5,
|
950 |
+
"learning_rate": 1.65149311484114e-05,
|
951 |
+
"loss": 0.0246,
|
952 |
+
"step": 119000
|
953 |
+
},
|
954 |
+
{
|
955 |
+
"epoch": 8.57,
|
956 |
+
"learning_rate": 1.500216365317587e-05,
|
957 |
+
"loss": 0.025,
|
958 |
+
"step": 120000
|
959 |
+
},
|
960 |
+
{
|
961 |
+
"epoch": 8.57,
|
962 |
+
"eval_cer": 0.09134037296954592,
|
963 |
+
"eval_loss": 0.045206133276224136,
|
964 |
+
"eval_runtime": 93.5201,
|
965 |
+
"eval_samples_per_second": 52.93,
|
966 |
+
"eval_steps_per_second": 0.417,
|
967 |
+
"eval_wer": 0.25191919191919193,
|
968 |
+
"step": 120000
|
969 |
+
},
|
970 |
+
{
|
971 |
+
"epoch": 8.65,
|
972 |
+
"learning_rate": 1.355840280886582e-05,
|
973 |
+
"loss": 0.025,
|
974 |
+
"step": 121000
|
975 |
+
},
|
976 |
+
{
|
977 |
+
"epoch": 8.72,
|
978 |
+
"learning_rate": 1.2184386620647097e-05,
|
979 |
+
"loss": 0.0247,
|
980 |
+
"step": 122000
|
981 |
+
},
|
982 |
+
{
|
983 |
+
"epoch": 8.79,
|
984 |
+
"learning_rate": 1.0880817442409478e-05,
|
985 |
+
"loss": 0.0248,
|
986 |
+
"step": 123000
|
987 |
+
},
|
988 |
+
{
|
989 |
+
"epoch": 8.86,
|
990 |
+
"learning_rate": 9.648361617745371e-06,
|
991 |
+
"loss": 0.0248,
|
992 |
+
"step": 124000
|
993 |
+
},
|
994 |
+
{
|
995 |
+
"epoch": 8.93,
|
996 |
+
"learning_rate": 8.487649139335962e-06,
|
997 |
+
"loss": 0.0248,
|
998 |
+
"step": 125000
|
999 |
+
},
|
1000 |
+
{
|
1001 |
+
"epoch": 8.93,
|
1002 |
+
"eval_cer": 0.09169016855774907,
|
1003 |
+
"eval_loss": 0.04524415731430054,
|
1004 |
+
"eval_runtime": 93.2177,
|
1005 |
+
"eval_samples_per_second": 53.101,
|
1006 |
+
"eval_steps_per_second": 0.418,
|
1007 |
+
"eval_wer": 0.25292929292929295,
|
1008 |
+
"step": 125000
|
1009 |
+
},
|
1010 |
+
{
|
1011 |
+
"epoch": 9.0,
|
1012 |
+
"learning_rate": 7.399273326918692e-06,
|
1013 |
+
"loss": 0.0246,
|
1014 |
+
"step": 126000
|
1015 |
+
},
|
1016 |
+
{
|
1017 |
+
"epoch": 9.07,
|
1018 |
+
"learning_rate": 6.383790524001009e-06,
|
1019 |
+
"loss": 0.024,
|
1020 |
+
"step": 127000
|
1021 |
+
},
|
1022 |
+
{
|
1023 |
+
"epoch": 9.15,
|
1024 |
+
"learning_rate": 5.441719813474849e-06,
|
1025 |
+
"loss": 0.0241,
|
1026 |
+
"step": 128000
|
1027 |
+
},
|
1028 |
+
{
|
1029 |
+
"epoch": 9.22,
|
1030 |
+
"learning_rate": 4.57354275227797e-06,
|
1031 |
+
"loss": 0.0242,
|
1032 |
+
"step": 129000
|
1033 |
+
},
|
1034 |
+
{
|
1035 |
+
"epoch": 9.29,
|
1036 |
+
"learning_rate": 3.7797031252369767e-06,
|
1037 |
+
"loss": 0.0243,
|
1038 |
+
"step": 130000
|
1039 |
+
},
|
1040 |
+
{
|
1041 |
+
"epoch": 9.29,
|
1042 |
+
"eval_cer": 0.09138409741807131,
|
1043 |
+
"eval_loss": 0.0453341044485569,
|
1044 |
+
"eval_runtime": 93.6155,
|
1045 |
+
"eval_samples_per_second": 52.876,
|
1046 |
+
"eval_steps_per_second": 0.417,
|
1047 |
+
"eval_wer": 0.25212121212121213,
|
1048 |
+
"step": 130000
|
1049 |
+
},
|
1050 |
+
{
|
1051 |
+
"epoch": 9.36,
|
1052 |
+
"learning_rate": 3.0606067182186776e-06,
|
1053 |
+
"loss": 0.024,
|
1054 |
+
"step": 131000
|
1055 |
+
},
|
1056 |
+
{
|
1057 |
+
"epoch": 9.43,
|
1058 |
+
"learning_rate": 2.4166211107049584e-06,
|
1059 |
+
"loss": 0.0242,
|
1060 |
+
"step": 132000
|
1061 |
+
},
|
1062 |
+
{
|
1063 |
+
"epoch": 9.5,
|
1064 |
+
"learning_rate": 1.8480754878977489e-06,
|
1065 |
+
"loss": 0.0241,
|
1066 |
+
"step": 133000
|
1067 |
+
},
|
1068 |
+
{
|
1069 |
+
"epoch": 9.57,
|
1070 |
+
"learning_rate": 1.3552604724498928e-06,
|
1071 |
+
"loss": 0.0241,
|
1072 |
+
"step": 134000
|
1073 |
+
},
|
1074 |
+
{
|
1075 |
+
"epoch": 9.65,
|
1076 |
+
"learning_rate": 9.384279759080127e-07,
|
1077 |
+
"loss": 0.024,
|
1078 |
+
"step": 135000
|
1079 |
+
},
|
1080 |
+
{
|
1081 |
+
"epoch": 9.65,
|
1082 |
+
"eval_cer": 0.09129664852102053,
|
1083 |
+
"eval_loss": 0.04531262069940567,
|
1084 |
+
"eval_runtime": 93.4097,
|
1085 |
+
"eval_samples_per_second": 52.992,
|
1086 |
+
"eval_steps_per_second": 0.418,
|
1087 |
+
"eval_wer": 0.25212121212121213,
|
1088 |
+
"step": 135000
|
1089 |
+
}
|
1090 |
+
],
|
1091 |
+
"max_steps": 139950,
|
1092 |
+
"num_train_epochs": 10,
|
1093 |
+
"total_flos": 4.19385405764928e+18,
|
1094 |
+
"trial_name": null,
|
1095 |
+
"trial_params": null
|
1096 |
+
}
|