End of training
Browse files- all_results.json +7 -0
- train_results.json +7 -0
- trainer_state.json +1033 -0
all_results.json
ADDED
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 60.0,
|
3 |
+
"train_loss": 0.18218991573040302,
|
4 |
+
"train_runtime": 3746.5818,
|
5 |
+
"train_samples_per_second": 39.876,
|
6 |
+
"train_steps_per_second": 2.498
|
7 |
+
}
|
train_results.json
ADDED
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 60.0,
|
3 |
+
"train_loss": 0.18218991573040302,
|
4 |
+
"train_runtime": 3746.5818,
|
5 |
+
"train_samples_per_second": 39.876,
|
6 |
+
"train_steps_per_second": 2.498
|
7 |
+
}
|
trainer_state.json
ADDED
@@ -0,0 +1,1033 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 60.0,
|
5 |
+
"global_step": 9360,
|
6 |
+
"is_hyper_param_search": false,
|
7 |
+
"is_local_process_zero": true,
|
8 |
+
"is_world_process_zero": true,
|
9 |
+
"log_history": [
|
10 |
+
{
|
11 |
+
"epoch": 1.0,
|
12 |
+
"eval_accuracy": 0.5270758122743683,
|
13 |
+
"eval_loss": 0.42201343178749084,
|
14 |
+
"eval_runtime": 4.2448,
|
15 |
+
"eval_samples_per_second": 65.256,
|
16 |
+
"eval_steps_per_second": 8.245,
|
17 |
+
"step": 156
|
18 |
+
},
|
19 |
+
{
|
20 |
+
"best_epoch": 0,
|
21 |
+
"best_eval_accuracy": 0.5270758122743683,
|
22 |
+
"epoch": 1.0,
|
23 |
+
"step": 156
|
24 |
+
},
|
25 |
+
{
|
26 |
+
"epoch": 2.0,
|
27 |
+
"eval_accuracy": 0.4729241877256318,
|
28 |
+
"eval_loss": 0.2767408788204193,
|
29 |
+
"eval_runtime": 4.3201,
|
30 |
+
"eval_samples_per_second": 64.12,
|
31 |
+
"eval_steps_per_second": 8.102,
|
32 |
+
"step": 312
|
33 |
+
},
|
34 |
+
{
|
35 |
+
"best_epoch": 0,
|
36 |
+
"best_eval_accuracy": 0.5270758122743683,
|
37 |
+
"epoch": 2.0,
|
38 |
+
"step": 312
|
39 |
+
},
|
40 |
+
{
|
41 |
+
"epoch": 3.0,
|
42 |
+
"eval_accuracy": 0.4729241877256318,
|
43 |
+
"eval_loss": 0.4345341622829437,
|
44 |
+
"eval_runtime": 4.3481,
|
45 |
+
"eval_samples_per_second": 63.706,
|
46 |
+
"eval_steps_per_second": 8.05,
|
47 |
+
"step": 468
|
48 |
+
},
|
49 |
+
{
|
50 |
+
"best_epoch": 0,
|
51 |
+
"best_eval_accuracy": 0.5270758122743683,
|
52 |
+
"epoch": 3.0,
|
53 |
+
"step": 468
|
54 |
+
},
|
55 |
+
{
|
56 |
+
"epoch": 3.21,
|
57 |
+
"learning_rate": 0.0028397435897435895,
|
58 |
+
"loss": 0.2507,
|
59 |
+
"step": 500
|
60 |
+
},
|
61 |
+
{
|
62 |
+
"epoch": 4.0,
|
63 |
+
"eval_accuracy": 0.5342960288808665,
|
64 |
+
"eval_loss": 0.20055121183395386,
|
65 |
+
"eval_runtime": 4.3672,
|
66 |
+
"eval_samples_per_second": 63.427,
|
67 |
+
"eval_steps_per_second": 8.014,
|
68 |
+
"step": 624
|
69 |
+
},
|
70 |
+
{
|
71 |
+
"best_epoch": 3,
|
72 |
+
"best_eval_accuracy": 0.5342960288808665,
|
73 |
+
"epoch": 4.0,
|
74 |
+
"step": 624
|
75 |
+
},
|
76 |
+
{
|
77 |
+
"epoch": 5.0,
|
78 |
+
"eval_accuracy": 0.4729241877256318,
|
79 |
+
"eval_loss": 0.17973236739635468,
|
80 |
+
"eval_runtime": 4.3589,
|
81 |
+
"eval_samples_per_second": 63.548,
|
82 |
+
"eval_steps_per_second": 8.029,
|
83 |
+
"step": 780
|
84 |
+
},
|
85 |
+
{
|
86 |
+
"best_epoch": 3,
|
87 |
+
"best_eval_accuracy": 0.5342960288808665,
|
88 |
+
"epoch": 5.0,
|
89 |
+
"step": 780
|
90 |
+
},
|
91 |
+
{
|
92 |
+
"epoch": 6.0,
|
93 |
+
"eval_accuracy": 0.5270758122743683,
|
94 |
+
"eval_loss": 0.21800895035266876,
|
95 |
+
"eval_runtime": 4.3571,
|
96 |
+
"eval_samples_per_second": 63.575,
|
97 |
+
"eval_steps_per_second": 8.033,
|
98 |
+
"step": 936
|
99 |
+
},
|
100 |
+
{
|
101 |
+
"best_epoch": 3,
|
102 |
+
"best_eval_accuracy": 0.5342960288808665,
|
103 |
+
"epoch": 6.0,
|
104 |
+
"step": 936
|
105 |
+
},
|
106 |
+
{
|
107 |
+
"epoch": 6.41,
|
108 |
+
"learning_rate": 0.00267948717948718,
|
109 |
+
"loss": 0.2023,
|
110 |
+
"step": 1000
|
111 |
+
},
|
112 |
+
{
|
113 |
+
"epoch": 7.0,
|
114 |
+
"eval_accuracy": 0.5054151624548736,
|
115 |
+
"eval_loss": 0.17262107133865356,
|
116 |
+
"eval_runtime": 4.3549,
|
117 |
+
"eval_samples_per_second": 63.606,
|
118 |
+
"eval_steps_per_second": 8.037,
|
119 |
+
"step": 1092
|
120 |
+
},
|
121 |
+
{
|
122 |
+
"best_epoch": 3,
|
123 |
+
"best_eval_accuracy": 0.5342960288808665,
|
124 |
+
"epoch": 7.0,
|
125 |
+
"step": 1092
|
126 |
+
},
|
127 |
+
{
|
128 |
+
"epoch": 8.0,
|
129 |
+
"eval_accuracy": 0.4729241877256318,
|
130 |
+
"eval_loss": 0.181132510304451,
|
131 |
+
"eval_runtime": 4.3482,
|
132 |
+
"eval_samples_per_second": 63.705,
|
133 |
+
"eval_steps_per_second": 8.049,
|
134 |
+
"step": 1248
|
135 |
+
},
|
136 |
+
{
|
137 |
+
"best_epoch": 3,
|
138 |
+
"best_eval_accuracy": 0.5342960288808665,
|
139 |
+
"epoch": 8.0,
|
140 |
+
"step": 1248
|
141 |
+
},
|
142 |
+
{
|
143 |
+
"epoch": 9.0,
|
144 |
+
"eval_accuracy": 0.5451263537906137,
|
145 |
+
"eval_loss": 0.1828141212463379,
|
146 |
+
"eval_runtime": 4.3504,
|
147 |
+
"eval_samples_per_second": 63.673,
|
148 |
+
"eval_steps_per_second": 8.045,
|
149 |
+
"step": 1404
|
150 |
+
},
|
151 |
+
{
|
152 |
+
"best_epoch": 8,
|
153 |
+
"best_eval_accuracy": 0.5451263537906137,
|
154 |
+
"epoch": 9.0,
|
155 |
+
"step": 1404
|
156 |
+
},
|
157 |
+
{
|
158 |
+
"epoch": 9.62,
|
159 |
+
"learning_rate": 0.0025192307692307693,
|
160 |
+
"loss": 0.2077,
|
161 |
+
"step": 1500
|
162 |
+
},
|
163 |
+
{
|
164 |
+
"epoch": 10.0,
|
165 |
+
"eval_accuracy": 0.5342960288808665,
|
166 |
+
"eval_loss": 0.19208335876464844,
|
167 |
+
"eval_runtime": 4.3537,
|
168 |
+
"eval_samples_per_second": 63.624,
|
169 |
+
"eval_steps_per_second": 8.039,
|
170 |
+
"step": 1560
|
171 |
+
},
|
172 |
+
{
|
173 |
+
"best_epoch": 8,
|
174 |
+
"best_eval_accuracy": 0.5451263537906137,
|
175 |
+
"epoch": 10.0,
|
176 |
+
"step": 1560
|
177 |
+
},
|
178 |
+
{
|
179 |
+
"epoch": 11.0,
|
180 |
+
"eval_accuracy": 0.48375451263537905,
|
181 |
+
"eval_loss": 0.17719601094722748,
|
182 |
+
"eval_runtime": 4.3563,
|
183 |
+
"eval_samples_per_second": 63.587,
|
184 |
+
"eval_steps_per_second": 8.034,
|
185 |
+
"step": 1716
|
186 |
+
},
|
187 |
+
{
|
188 |
+
"best_epoch": 8,
|
189 |
+
"best_eval_accuracy": 0.5451263537906137,
|
190 |
+
"epoch": 11.0,
|
191 |
+
"step": 1716
|
192 |
+
},
|
193 |
+
{
|
194 |
+
"epoch": 12.0,
|
195 |
+
"eval_accuracy": 0.6462093862815884,
|
196 |
+
"eval_loss": 0.17240847647190094,
|
197 |
+
"eval_runtime": 4.3552,
|
198 |
+
"eval_samples_per_second": 63.602,
|
199 |
+
"eval_steps_per_second": 8.036,
|
200 |
+
"step": 1872
|
201 |
+
},
|
202 |
+
{
|
203 |
+
"best_epoch": 11,
|
204 |
+
"best_eval_accuracy": 0.6462093862815884,
|
205 |
+
"epoch": 12.0,
|
206 |
+
"step": 1872
|
207 |
+
},
|
208 |
+
{
|
209 |
+
"epoch": 12.82,
|
210 |
+
"learning_rate": 0.002358974358974359,
|
211 |
+
"loss": 0.189,
|
212 |
+
"step": 2000
|
213 |
+
},
|
214 |
+
{
|
215 |
+
"epoch": 13.0,
|
216 |
+
"eval_accuracy": 0.5379061371841155,
|
217 |
+
"eval_loss": 0.17178961634635925,
|
218 |
+
"eval_runtime": 4.3528,
|
219 |
+
"eval_samples_per_second": 63.637,
|
220 |
+
"eval_steps_per_second": 8.041,
|
221 |
+
"step": 2028
|
222 |
+
},
|
223 |
+
{
|
224 |
+
"best_epoch": 11,
|
225 |
+
"best_eval_accuracy": 0.6462093862815884,
|
226 |
+
"epoch": 13.0,
|
227 |
+
"step": 2028
|
228 |
+
},
|
229 |
+
{
|
230 |
+
"epoch": 14.0,
|
231 |
+
"eval_accuracy": 0.5126353790613718,
|
232 |
+
"eval_loss": 0.1727617383003235,
|
233 |
+
"eval_runtime": 4.352,
|
234 |
+
"eval_samples_per_second": 63.648,
|
235 |
+
"eval_steps_per_second": 8.042,
|
236 |
+
"step": 2184
|
237 |
+
},
|
238 |
+
{
|
239 |
+
"best_epoch": 11,
|
240 |
+
"best_eval_accuracy": 0.6462093862815884,
|
241 |
+
"epoch": 14.0,
|
242 |
+
"step": 2184
|
243 |
+
},
|
244 |
+
{
|
245 |
+
"epoch": 15.0,
|
246 |
+
"eval_accuracy": 0.5126353790613718,
|
247 |
+
"eval_loss": 0.17747661471366882,
|
248 |
+
"eval_runtime": 4.357,
|
249 |
+
"eval_samples_per_second": 63.576,
|
250 |
+
"eval_steps_per_second": 8.033,
|
251 |
+
"step": 2340
|
252 |
+
},
|
253 |
+
{
|
254 |
+
"best_epoch": 11,
|
255 |
+
"best_eval_accuracy": 0.6462093862815884,
|
256 |
+
"epoch": 15.0,
|
257 |
+
"step": 2340
|
258 |
+
},
|
259 |
+
{
|
260 |
+
"epoch": 16.0,
|
261 |
+
"eval_accuracy": 0.5595667870036101,
|
262 |
+
"eval_loss": 0.18130473792552948,
|
263 |
+
"eval_runtime": 4.3517,
|
264 |
+
"eval_samples_per_second": 63.654,
|
265 |
+
"eval_steps_per_second": 8.043,
|
266 |
+
"step": 2496
|
267 |
+
},
|
268 |
+
{
|
269 |
+
"best_epoch": 11,
|
270 |
+
"best_eval_accuracy": 0.6462093862815884,
|
271 |
+
"epoch": 16.0,
|
272 |
+
"step": 2496
|
273 |
+
},
|
274 |
+
{
|
275 |
+
"epoch": 16.03,
|
276 |
+
"learning_rate": 0.0021987179487179486,
|
277 |
+
"loss": 0.1803,
|
278 |
+
"step": 2500
|
279 |
+
},
|
280 |
+
{
|
281 |
+
"epoch": 17.0,
|
282 |
+
"eval_accuracy": 0.631768953068592,
|
283 |
+
"eval_loss": 0.17390955984592438,
|
284 |
+
"eval_runtime": 4.3538,
|
285 |
+
"eval_samples_per_second": 63.623,
|
286 |
+
"eval_steps_per_second": 8.039,
|
287 |
+
"step": 2652
|
288 |
+
},
|
289 |
+
{
|
290 |
+
"best_epoch": 11,
|
291 |
+
"best_eval_accuracy": 0.6462093862815884,
|
292 |
+
"epoch": 17.0,
|
293 |
+
"step": 2652
|
294 |
+
},
|
295 |
+
{
|
296 |
+
"epoch": 18.0,
|
297 |
+
"eval_accuracy": 0.6137184115523465,
|
298 |
+
"eval_loss": 0.17176125943660736,
|
299 |
+
"eval_runtime": 4.3563,
|
300 |
+
"eval_samples_per_second": 63.586,
|
301 |
+
"eval_steps_per_second": 8.034,
|
302 |
+
"step": 2808
|
303 |
+
},
|
304 |
+
{
|
305 |
+
"best_epoch": 11,
|
306 |
+
"best_eval_accuracy": 0.6462093862815884,
|
307 |
+
"epoch": 18.0,
|
308 |
+
"step": 2808
|
309 |
+
},
|
310 |
+
{
|
311 |
+
"epoch": 19.0,
|
312 |
+
"eval_accuracy": 0.6389891696750902,
|
313 |
+
"eval_loss": 0.1711309403181076,
|
314 |
+
"eval_runtime": 4.3659,
|
315 |
+
"eval_samples_per_second": 63.446,
|
316 |
+
"eval_steps_per_second": 8.017,
|
317 |
+
"step": 2964
|
318 |
+
},
|
319 |
+
{
|
320 |
+
"best_epoch": 11,
|
321 |
+
"best_eval_accuracy": 0.6462093862815884,
|
322 |
+
"epoch": 19.0,
|
323 |
+
"step": 2964
|
324 |
+
},
|
325 |
+
{
|
326 |
+
"epoch": 19.23,
|
327 |
+
"learning_rate": 0.0020384615384615385,
|
328 |
+
"loss": 0.1791,
|
329 |
+
"step": 3000
|
330 |
+
},
|
331 |
+
{
|
332 |
+
"epoch": 20.0,
|
333 |
+
"eval_accuracy": 0.5956678700361011,
|
334 |
+
"eval_loss": 0.179701030254364,
|
335 |
+
"eval_runtime": 4.3561,
|
336 |
+
"eval_samples_per_second": 63.589,
|
337 |
+
"eval_steps_per_second": 8.035,
|
338 |
+
"step": 3120
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"best_epoch": 11,
|
342 |
+
"best_eval_accuracy": 0.6462093862815884,
|
343 |
+
"epoch": 20.0,
|
344 |
+
"step": 3120
|
345 |
+
},
|
346 |
+
{
|
347 |
+
"epoch": 21.0,
|
348 |
+
"eval_accuracy": 0.6859205776173285,
|
349 |
+
"eval_loss": 0.17101024091243744,
|
350 |
+
"eval_runtime": 4.3528,
|
351 |
+
"eval_samples_per_second": 63.637,
|
352 |
+
"eval_steps_per_second": 8.041,
|
353 |
+
"step": 3276
|
354 |
+
},
|
355 |
+
{
|
356 |
+
"best_epoch": 20,
|
357 |
+
"best_eval_accuracy": 0.6859205776173285,
|
358 |
+
"epoch": 21.0,
|
359 |
+
"step": 3276
|
360 |
+
},
|
361 |
+
{
|
362 |
+
"epoch": 22.0,
|
363 |
+
"eval_accuracy": 0.6642599277978339,
|
364 |
+
"eval_loss": 0.17285466194152832,
|
365 |
+
"eval_runtime": 4.3557,
|
366 |
+
"eval_samples_per_second": 63.595,
|
367 |
+
"eval_steps_per_second": 8.035,
|
368 |
+
"step": 3432
|
369 |
+
},
|
370 |
+
{
|
371 |
+
"best_epoch": 20,
|
372 |
+
"best_eval_accuracy": 0.6859205776173285,
|
373 |
+
"epoch": 22.0,
|
374 |
+
"step": 3432
|
375 |
+
},
|
376 |
+
{
|
377 |
+
"epoch": 22.44,
|
378 |
+
"learning_rate": 0.0018782051282051281,
|
379 |
+
"loss": 0.1781,
|
380 |
+
"step": 3500
|
381 |
+
},
|
382 |
+
{
|
383 |
+
"epoch": 23.0,
|
384 |
+
"eval_accuracy": 0.6823104693140795,
|
385 |
+
"eval_loss": 0.1700609028339386,
|
386 |
+
"eval_runtime": 4.345,
|
387 |
+
"eval_samples_per_second": 63.751,
|
388 |
+
"eval_steps_per_second": 8.055,
|
389 |
+
"step": 3588
|
390 |
+
},
|
391 |
+
{
|
392 |
+
"best_epoch": 20,
|
393 |
+
"best_eval_accuracy": 0.6859205776173285,
|
394 |
+
"epoch": 23.0,
|
395 |
+
"step": 3588
|
396 |
+
},
|
397 |
+
{
|
398 |
+
"epoch": 24.0,
|
399 |
+
"eval_accuracy": 0.6389891696750902,
|
400 |
+
"eval_loss": 0.17063240706920624,
|
401 |
+
"eval_runtime": 4.3466,
|
402 |
+
"eval_samples_per_second": 63.728,
|
403 |
+
"eval_steps_per_second": 8.052,
|
404 |
+
"step": 3744
|
405 |
+
},
|
406 |
+
{
|
407 |
+
"best_epoch": 20,
|
408 |
+
"best_eval_accuracy": 0.6859205776173285,
|
409 |
+
"epoch": 24.0,
|
410 |
+
"step": 3744
|
411 |
+
},
|
412 |
+
{
|
413 |
+
"epoch": 25.0,
|
414 |
+
"eval_accuracy": 0.6859205776173285,
|
415 |
+
"eval_loss": 0.17078790068626404,
|
416 |
+
"eval_runtime": 4.3477,
|
417 |
+
"eval_samples_per_second": 63.711,
|
418 |
+
"eval_steps_per_second": 8.05,
|
419 |
+
"step": 3900
|
420 |
+
},
|
421 |
+
{
|
422 |
+
"best_epoch": 20,
|
423 |
+
"best_eval_accuracy": 0.6859205776173285,
|
424 |
+
"epoch": 25.0,
|
425 |
+
"step": 3900
|
426 |
+
},
|
427 |
+
{
|
428 |
+
"epoch": 25.64,
|
429 |
+
"learning_rate": 0.0017179487179487178,
|
430 |
+
"loss": 0.1765,
|
431 |
+
"step": 4000
|
432 |
+
},
|
433 |
+
{
|
434 |
+
"epoch": 26.0,
|
435 |
+
"eval_accuracy": 0.6642599277978339,
|
436 |
+
"eval_loss": 0.1696767657995224,
|
437 |
+
"eval_runtime": 4.363,
|
438 |
+
"eval_samples_per_second": 63.488,
|
439 |
+
"eval_steps_per_second": 8.022,
|
440 |
+
"step": 4056
|
441 |
+
},
|
442 |
+
{
|
443 |
+
"best_epoch": 20,
|
444 |
+
"best_eval_accuracy": 0.6859205776173285,
|
445 |
+
"epoch": 26.0,
|
446 |
+
"step": 4056
|
447 |
+
},
|
448 |
+
{
|
449 |
+
"epoch": 27.0,
|
450 |
+
"eval_accuracy": 0.6714801444043321,
|
451 |
+
"eval_loss": 0.16979683935642242,
|
452 |
+
"eval_runtime": 4.3432,
|
453 |
+
"eval_samples_per_second": 63.778,
|
454 |
+
"eval_steps_per_second": 8.059,
|
455 |
+
"step": 4212
|
456 |
+
},
|
457 |
+
{
|
458 |
+
"best_epoch": 20,
|
459 |
+
"best_eval_accuracy": 0.6859205776173285,
|
460 |
+
"epoch": 27.0,
|
461 |
+
"step": 4212
|
462 |
+
},
|
463 |
+
{
|
464 |
+
"epoch": 28.0,
|
465 |
+
"eval_accuracy": 0.6425992779783394,
|
466 |
+
"eval_loss": 0.17104002833366394,
|
467 |
+
"eval_runtime": 4.3513,
|
468 |
+
"eval_samples_per_second": 63.66,
|
469 |
+
"eval_steps_per_second": 8.044,
|
470 |
+
"step": 4368
|
471 |
+
},
|
472 |
+
{
|
473 |
+
"best_epoch": 20,
|
474 |
+
"best_eval_accuracy": 0.6859205776173285,
|
475 |
+
"epoch": 28.0,
|
476 |
+
"step": 4368
|
477 |
+
},
|
478 |
+
{
|
479 |
+
"epoch": 28.85,
|
480 |
+
"learning_rate": 0.0015576923076923079,
|
481 |
+
"loss": 0.176,
|
482 |
+
"step": 4500
|
483 |
+
},
|
484 |
+
{
|
485 |
+
"epoch": 29.0,
|
486 |
+
"eval_accuracy": 0.6931407942238267,
|
487 |
+
"eval_loss": 0.17095446586608887,
|
488 |
+
"eval_runtime": 4.3498,
|
489 |
+
"eval_samples_per_second": 63.681,
|
490 |
+
"eval_steps_per_second": 8.046,
|
491 |
+
"step": 4524
|
492 |
+
},
|
493 |
+
{
|
494 |
+
"best_epoch": 28,
|
495 |
+
"best_eval_accuracy": 0.6931407942238267,
|
496 |
+
"epoch": 29.0,
|
497 |
+
"step": 4524
|
498 |
+
},
|
499 |
+
{
|
500 |
+
"epoch": 30.0,
|
501 |
+
"eval_accuracy": 0.6967509025270758,
|
502 |
+
"eval_loss": 0.1702578067779541,
|
503 |
+
"eval_runtime": 4.3424,
|
504 |
+
"eval_samples_per_second": 63.79,
|
505 |
+
"eval_steps_per_second": 8.06,
|
506 |
+
"step": 4680
|
507 |
+
},
|
508 |
+
{
|
509 |
+
"best_epoch": 29,
|
510 |
+
"best_eval_accuracy": 0.6967509025270758,
|
511 |
+
"epoch": 30.0,
|
512 |
+
"step": 4680
|
513 |
+
},
|
514 |
+
{
|
515 |
+
"epoch": 31.0,
|
516 |
+
"eval_accuracy": 0.6570397111913358,
|
517 |
+
"eval_loss": 0.17254619300365448,
|
518 |
+
"eval_runtime": 4.3471,
|
519 |
+
"eval_samples_per_second": 63.721,
|
520 |
+
"eval_steps_per_second": 8.051,
|
521 |
+
"step": 4836
|
522 |
+
},
|
523 |
+
{
|
524 |
+
"best_epoch": 29,
|
525 |
+
"best_eval_accuracy": 0.6967509025270758,
|
526 |
+
"epoch": 31.0,
|
527 |
+
"step": 4836
|
528 |
+
},
|
529 |
+
{
|
530 |
+
"epoch": 32.0,
|
531 |
+
"eval_accuracy": 0.6714801444043321,
|
532 |
+
"eval_loss": 0.1699478179216385,
|
533 |
+
"eval_runtime": 4.3442,
|
534 |
+
"eval_samples_per_second": 63.763,
|
535 |
+
"eval_steps_per_second": 8.057,
|
536 |
+
"step": 4992
|
537 |
+
},
|
538 |
+
{
|
539 |
+
"best_epoch": 29,
|
540 |
+
"best_eval_accuracy": 0.6967509025270758,
|
541 |
+
"epoch": 32.0,
|
542 |
+
"step": 4992
|
543 |
+
},
|
544 |
+
{
|
545 |
+
"epoch": 32.05,
|
546 |
+
"learning_rate": 0.0013974358974358976,
|
547 |
+
"loss": 0.1749,
|
548 |
+
"step": 5000
|
549 |
+
},
|
550 |
+
{
|
551 |
+
"epoch": 33.0,
|
552 |
+
"eval_accuracy": 0.6895306859205776,
|
553 |
+
"eval_loss": 0.17102044820785522,
|
554 |
+
"eval_runtime": 4.3501,
|
555 |
+
"eval_samples_per_second": 63.677,
|
556 |
+
"eval_steps_per_second": 8.046,
|
557 |
+
"step": 5148
|
558 |
+
},
|
559 |
+
{
|
560 |
+
"best_epoch": 29,
|
561 |
+
"best_eval_accuracy": 0.6967509025270758,
|
562 |
+
"epoch": 33.0,
|
563 |
+
"step": 5148
|
564 |
+
},
|
565 |
+
{
|
566 |
+
"epoch": 34.0,
|
567 |
+
"eval_accuracy": 0.7220216606498195,
|
568 |
+
"eval_loss": 0.1694263070821762,
|
569 |
+
"eval_runtime": 4.347,
|
570 |
+
"eval_samples_per_second": 63.722,
|
571 |
+
"eval_steps_per_second": 8.051,
|
572 |
+
"step": 5304
|
573 |
+
},
|
574 |
+
{
|
575 |
+
"best_epoch": 33,
|
576 |
+
"best_eval_accuracy": 0.7220216606498195,
|
577 |
+
"epoch": 34.0,
|
578 |
+
"step": 5304
|
579 |
+
},
|
580 |
+
{
|
581 |
+
"epoch": 35.0,
|
582 |
+
"eval_accuracy": 0.6534296028880866,
|
583 |
+
"eval_loss": 0.16998881101608276,
|
584 |
+
"eval_runtime": 4.3466,
|
585 |
+
"eval_samples_per_second": 63.727,
|
586 |
+
"eval_steps_per_second": 8.052,
|
587 |
+
"step": 5460
|
588 |
+
},
|
589 |
+
{
|
590 |
+
"best_epoch": 33,
|
591 |
+
"best_eval_accuracy": 0.7220216606498195,
|
592 |
+
"epoch": 35.0,
|
593 |
+
"step": 5460
|
594 |
+
},
|
595 |
+
{
|
596 |
+
"epoch": 35.26,
|
597 |
+
"learning_rate": 0.0012371794871794872,
|
598 |
+
"loss": 0.1739,
|
599 |
+
"step": 5500
|
600 |
+
},
|
601 |
+
{
|
602 |
+
"epoch": 36.0,
|
603 |
+
"eval_accuracy": 0.7111913357400722,
|
604 |
+
"eval_loss": 0.16898111999034882,
|
605 |
+
"eval_runtime": 4.345,
|
606 |
+
"eval_samples_per_second": 63.751,
|
607 |
+
"eval_steps_per_second": 8.055,
|
608 |
+
"step": 5616
|
609 |
+
},
|
610 |
+
{
|
611 |
+
"best_epoch": 33,
|
612 |
+
"best_eval_accuracy": 0.7220216606498195,
|
613 |
+
"epoch": 36.0,
|
614 |
+
"step": 5616
|
615 |
+
},
|
616 |
+
{
|
617 |
+
"epoch": 37.0,
|
618 |
+
"eval_accuracy": 0.7220216606498195,
|
619 |
+
"eval_loss": 0.1684962809085846,
|
620 |
+
"eval_runtime": 4.3495,
|
621 |
+
"eval_samples_per_second": 63.686,
|
622 |
+
"eval_steps_per_second": 8.047,
|
623 |
+
"step": 5772
|
624 |
+
},
|
625 |
+
{
|
626 |
+
"best_epoch": 33,
|
627 |
+
"best_eval_accuracy": 0.7220216606498195,
|
628 |
+
"epoch": 37.0,
|
629 |
+
"step": 5772
|
630 |
+
},
|
631 |
+
{
|
632 |
+
"epoch": 38.0,
|
633 |
+
"eval_accuracy": 0.703971119133574,
|
634 |
+
"eval_loss": 0.1695842742919922,
|
635 |
+
"eval_runtime": 4.3499,
|
636 |
+
"eval_samples_per_second": 63.679,
|
637 |
+
"eval_steps_per_second": 8.046,
|
638 |
+
"step": 5928
|
639 |
+
},
|
640 |
+
{
|
641 |
+
"best_epoch": 33,
|
642 |
+
"best_eval_accuracy": 0.7220216606498195,
|
643 |
+
"epoch": 38.0,
|
644 |
+
"step": 5928
|
645 |
+
},
|
646 |
+
{
|
647 |
+
"epoch": 38.46,
|
648 |
+
"learning_rate": 0.0010769230769230769,
|
649 |
+
"loss": 0.1738,
|
650 |
+
"step": 6000
|
651 |
+
},
|
652 |
+
{
|
653 |
+
"epoch": 39.0,
|
654 |
+
"eval_accuracy": 0.7148014440433214,
|
655 |
+
"eval_loss": 0.16876961290836334,
|
656 |
+
"eval_runtime": 4.3498,
|
657 |
+
"eval_samples_per_second": 63.682,
|
658 |
+
"eval_steps_per_second": 8.046,
|
659 |
+
"step": 6084
|
660 |
+
},
|
661 |
+
{
|
662 |
+
"best_epoch": 33,
|
663 |
+
"best_eval_accuracy": 0.7220216606498195,
|
664 |
+
"epoch": 39.0,
|
665 |
+
"step": 6084
|
666 |
+
},
|
667 |
+
{
|
668 |
+
"epoch": 40.0,
|
669 |
+
"eval_accuracy": 0.7220216606498195,
|
670 |
+
"eval_loss": 0.16915743052959442,
|
671 |
+
"eval_runtime": 4.3454,
|
672 |
+
"eval_samples_per_second": 63.746,
|
673 |
+
"eval_steps_per_second": 8.055,
|
674 |
+
"step": 6240
|
675 |
+
},
|
676 |
+
{
|
677 |
+
"best_epoch": 33,
|
678 |
+
"best_eval_accuracy": 0.7220216606498195,
|
679 |
+
"epoch": 40.0,
|
680 |
+
"step": 6240
|
681 |
+
},
|
682 |
+
{
|
683 |
+
"epoch": 41.0,
|
684 |
+
"eval_accuracy": 0.7364620938628159,
|
685 |
+
"eval_loss": 0.16826777160167694,
|
686 |
+
"eval_runtime": 4.3504,
|
687 |
+
"eval_samples_per_second": 63.672,
|
688 |
+
"eval_steps_per_second": 8.045,
|
689 |
+
"step": 6396
|
690 |
+
},
|
691 |
+
{
|
692 |
+
"best_epoch": 40,
|
693 |
+
"best_eval_accuracy": 0.7364620938628159,
|
694 |
+
"epoch": 41.0,
|
695 |
+
"step": 6396
|
696 |
+
},
|
697 |
+
{
|
698 |
+
"epoch": 41.67,
|
699 |
+
"learning_rate": 0.0009166666666666668,
|
700 |
+
"loss": 0.1726,
|
701 |
+
"step": 6500
|
702 |
+
},
|
703 |
+
{
|
704 |
+
"epoch": 42.0,
|
705 |
+
"eval_accuracy": 0.6678700361010831,
|
706 |
+
"eval_loss": 0.16903138160705566,
|
707 |
+
"eval_runtime": 4.3443,
|
708 |
+
"eval_samples_per_second": 63.761,
|
709 |
+
"eval_steps_per_second": 8.056,
|
710 |
+
"step": 6552
|
711 |
+
},
|
712 |
+
{
|
713 |
+
"best_epoch": 40,
|
714 |
+
"best_eval_accuracy": 0.7364620938628159,
|
715 |
+
"epoch": 42.0,
|
716 |
+
"step": 6552
|
717 |
+
},
|
718 |
+
{
|
719 |
+
"epoch": 43.0,
|
720 |
+
"eval_accuracy": 0.7075812274368231,
|
721 |
+
"eval_loss": 0.16787345707416534,
|
722 |
+
"eval_runtime": 4.3441,
|
723 |
+
"eval_samples_per_second": 63.765,
|
724 |
+
"eval_steps_per_second": 8.057,
|
725 |
+
"step": 6708
|
726 |
+
},
|
727 |
+
{
|
728 |
+
"best_epoch": 40,
|
729 |
+
"best_eval_accuracy": 0.7364620938628159,
|
730 |
+
"epoch": 43.0,
|
731 |
+
"step": 6708
|
732 |
+
},
|
733 |
+
{
|
734 |
+
"epoch": 44.0,
|
735 |
+
"eval_accuracy": 0.7184115523465704,
|
736 |
+
"eval_loss": 0.16910089552402496,
|
737 |
+
"eval_runtime": 4.3519,
|
738 |
+
"eval_samples_per_second": 63.65,
|
739 |
+
"eval_steps_per_second": 8.042,
|
740 |
+
"step": 6864
|
741 |
+
},
|
742 |
+
{
|
743 |
+
"best_epoch": 40,
|
744 |
+
"best_eval_accuracy": 0.7364620938628159,
|
745 |
+
"epoch": 44.0,
|
746 |
+
"step": 6864
|
747 |
+
},
|
748 |
+
{
|
749 |
+
"epoch": 44.87,
|
750 |
+
"learning_rate": 0.0007564102564102564,
|
751 |
+
"loss": 0.1719,
|
752 |
+
"step": 7000
|
753 |
+
},
|
754 |
+
{
|
755 |
+
"epoch": 45.0,
|
756 |
+
"eval_accuracy": 0.7292418772563177,
|
757 |
+
"eval_loss": 0.16922158002853394,
|
758 |
+
"eval_runtime": 4.3511,
|
759 |
+
"eval_samples_per_second": 63.662,
|
760 |
+
"eval_steps_per_second": 8.044,
|
761 |
+
"step": 7020
|
762 |
+
},
|
763 |
+
{
|
764 |
+
"best_epoch": 40,
|
765 |
+
"best_eval_accuracy": 0.7364620938628159,
|
766 |
+
"epoch": 45.0,
|
767 |
+
"step": 7020
|
768 |
+
},
|
769 |
+
{
|
770 |
+
"epoch": 46.0,
|
771 |
+
"eval_accuracy": 0.7328519855595668,
|
772 |
+
"eval_loss": 0.16845819354057312,
|
773 |
+
"eval_runtime": 4.3526,
|
774 |
+
"eval_samples_per_second": 63.64,
|
775 |
+
"eval_steps_per_second": 8.041,
|
776 |
+
"step": 7176
|
777 |
+
},
|
778 |
+
{
|
779 |
+
"best_epoch": 40,
|
780 |
+
"best_eval_accuracy": 0.7364620938628159,
|
781 |
+
"epoch": 46.0,
|
782 |
+
"step": 7176
|
783 |
+
},
|
784 |
+
{
|
785 |
+
"epoch": 47.0,
|
786 |
+
"eval_accuracy": 0.7184115523465704,
|
787 |
+
"eval_loss": 0.1684226542711258,
|
788 |
+
"eval_runtime": 4.354,
|
789 |
+
"eval_samples_per_second": 63.619,
|
790 |
+
"eval_steps_per_second": 8.039,
|
791 |
+
"step": 7332
|
792 |
+
},
|
793 |
+
{
|
794 |
+
"best_epoch": 40,
|
795 |
+
"best_eval_accuracy": 0.7364620938628159,
|
796 |
+
"epoch": 47.0,
|
797 |
+
"step": 7332
|
798 |
+
},
|
799 |
+
{
|
800 |
+
"epoch": 48.0,
|
801 |
+
"eval_accuracy": 0.7111913357400722,
|
802 |
+
"eval_loss": 0.16897501051425934,
|
803 |
+
"eval_runtime": 4.3558,
|
804 |
+
"eval_samples_per_second": 63.593,
|
805 |
+
"eval_steps_per_second": 8.035,
|
806 |
+
"step": 7488
|
807 |
+
},
|
808 |
+
{
|
809 |
+
"best_epoch": 40,
|
810 |
+
"best_eval_accuracy": 0.7364620938628159,
|
811 |
+
"epoch": 48.0,
|
812 |
+
"step": 7488
|
813 |
+
},
|
814 |
+
{
|
815 |
+
"epoch": 48.08,
|
816 |
+
"learning_rate": 0.0005961538461538461,
|
817 |
+
"loss": 0.1712,
|
818 |
+
"step": 7500
|
819 |
+
},
|
820 |
+
{
|
821 |
+
"epoch": 49.0,
|
822 |
+
"eval_accuracy": 0.7292418772563177,
|
823 |
+
"eval_loss": 0.16902127861976624,
|
824 |
+
"eval_runtime": 4.3563,
|
825 |
+
"eval_samples_per_second": 63.585,
|
826 |
+
"eval_steps_per_second": 8.034,
|
827 |
+
"step": 7644
|
828 |
+
},
|
829 |
+
{
|
830 |
+
"best_epoch": 40,
|
831 |
+
"best_eval_accuracy": 0.7364620938628159,
|
832 |
+
"epoch": 49.0,
|
833 |
+
"step": 7644
|
834 |
+
},
|
835 |
+
{
|
836 |
+
"epoch": 50.0,
|
837 |
+
"eval_accuracy": 0.6931407942238267,
|
838 |
+
"eval_loss": 0.16854658722877502,
|
839 |
+
"eval_runtime": 4.3523,
|
840 |
+
"eval_samples_per_second": 63.645,
|
841 |
+
"eval_steps_per_second": 8.042,
|
842 |
+
"step": 7800
|
843 |
+
},
|
844 |
+
{
|
845 |
+
"best_epoch": 40,
|
846 |
+
"best_eval_accuracy": 0.7364620938628159,
|
847 |
+
"epoch": 50.0,
|
848 |
+
"step": 7800
|
849 |
+
},
|
850 |
+
{
|
851 |
+
"epoch": 51.0,
|
852 |
+
"eval_accuracy": 0.7256317689530686,
|
853 |
+
"eval_loss": 0.1680402308702469,
|
854 |
+
"eval_runtime": 4.3572,
|
855 |
+
"eval_samples_per_second": 63.573,
|
856 |
+
"eval_steps_per_second": 8.033,
|
857 |
+
"step": 7956
|
858 |
+
},
|
859 |
+
{
|
860 |
+
"best_epoch": 40,
|
861 |
+
"best_eval_accuracy": 0.7364620938628159,
|
862 |
+
"epoch": 51.0,
|
863 |
+
"step": 7956
|
864 |
+
},
|
865 |
+
{
|
866 |
+
"epoch": 51.28,
|
867 |
+
"learning_rate": 0.00043589743589743596,
|
868 |
+
"loss": 0.1705,
|
869 |
+
"step": 8000
|
870 |
+
},
|
871 |
+
{
|
872 |
+
"epoch": 52.0,
|
873 |
+
"eval_accuracy": 0.7075812274368231,
|
874 |
+
"eval_loss": 0.1686880886554718,
|
875 |
+
"eval_runtime": 4.355,
|
876 |
+
"eval_samples_per_second": 63.606,
|
877 |
+
"eval_steps_per_second": 8.037,
|
878 |
+
"step": 8112
|
879 |
+
},
|
880 |
+
{
|
881 |
+
"best_epoch": 40,
|
882 |
+
"best_eval_accuracy": 0.7364620938628159,
|
883 |
+
"epoch": 52.0,
|
884 |
+
"step": 8112
|
885 |
+
},
|
886 |
+
{
|
887 |
+
"epoch": 53.0,
|
888 |
+
"eval_accuracy": 0.7184115523465704,
|
889 |
+
"eval_loss": 0.1684812754392624,
|
890 |
+
"eval_runtime": 4.3562,
|
891 |
+
"eval_samples_per_second": 63.587,
|
892 |
+
"eval_steps_per_second": 8.034,
|
893 |
+
"step": 8268
|
894 |
+
},
|
895 |
+
{
|
896 |
+
"best_epoch": 40,
|
897 |
+
"best_eval_accuracy": 0.7364620938628159,
|
898 |
+
"epoch": 53.0,
|
899 |
+
"step": 8268
|
900 |
+
},
|
901 |
+
{
|
902 |
+
"epoch": 54.0,
|
903 |
+
"eval_accuracy": 0.7364620938628159,
|
904 |
+
"eval_loss": 0.1689344346523285,
|
905 |
+
"eval_runtime": 4.3562,
|
906 |
+
"eval_samples_per_second": 63.587,
|
907 |
+
"eval_steps_per_second": 8.034,
|
908 |
+
"step": 8424
|
909 |
+
},
|
910 |
+
{
|
911 |
+
"best_epoch": 40,
|
912 |
+
"best_eval_accuracy": 0.7364620938628159,
|
913 |
+
"epoch": 54.0,
|
914 |
+
"step": 8424
|
915 |
+
},
|
916 |
+
{
|
917 |
+
"epoch": 54.49,
|
918 |
+
"learning_rate": 0.0002756410256410257,
|
919 |
+
"loss": 0.1705,
|
920 |
+
"step": 8500
|
921 |
+
},
|
922 |
+
{
|
923 |
+
"epoch": 55.0,
|
924 |
+
"eval_accuracy": 0.7148014440433214,
|
925 |
+
"eval_loss": 0.16774678230285645,
|
926 |
+
"eval_runtime": 4.355,
|
927 |
+
"eval_samples_per_second": 63.606,
|
928 |
+
"eval_steps_per_second": 8.037,
|
929 |
+
"step": 8580
|
930 |
+
},
|
931 |
+
{
|
932 |
+
"best_epoch": 40,
|
933 |
+
"best_eval_accuracy": 0.7364620938628159,
|
934 |
+
"epoch": 55.0,
|
935 |
+
"step": 8580
|
936 |
+
},
|
937 |
+
{
|
938 |
+
"epoch": 56.0,
|
939 |
+
"eval_accuracy": 0.7220216606498195,
|
940 |
+
"eval_loss": 0.1693679541349411,
|
941 |
+
"eval_runtime": 4.3485,
|
942 |
+
"eval_samples_per_second": 63.7,
|
943 |
+
"eval_steps_per_second": 8.049,
|
944 |
+
"step": 8736
|
945 |
+
},
|
946 |
+
{
|
947 |
+
"best_epoch": 40,
|
948 |
+
"best_eval_accuracy": 0.7364620938628159,
|
949 |
+
"epoch": 56.0,
|
950 |
+
"step": 8736
|
951 |
+
},
|
952 |
+
{
|
953 |
+
"epoch": 57.0,
|
954 |
+
"eval_accuracy": 0.7256317689530686,
|
955 |
+
"eval_loss": 0.16823026537895203,
|
956 |
+
"eval_runtime": 4.3527,
|
957 |
+
"eval_samples_per_second": 63.639,
|
958 |
+
"eval_steps_per_second": 8.041,
|
959 |
+
"step": 8892
|
960 |
+
},
|
961 |
+
{
|
962 |
+
"best_epoch": 40,
|
963 |
+
"best_eval_accuracy": 0.7364620938628159,
|
964 |
+
"epoch": 57.0,
|
965 |
+
"step": 8892
|
966 |
+
},
|
967 |
+
{
|
968 |
+
"epoch": 57.69,
|
969 |
+
"learning_rate": 0.0001153846153846154,
|
970 |
+
"loss": 0.1692,
|
971 |
+
"step": 9000
|
972 |
+
},
|
973 |
+
{
|
974 |
+
"epoch": 58.0,
|
975 |
+
"eval_accuracy": 0.7148014440433214,
|
976 |
+
"eval_loss": 0.16843324899673462,
|
977 |
+
"eval_runtime": 4.348,
|
978 |
+
"eval_samples_per_second": 63.707,
|
979 |
+
"eval_steps_per_second": 8.05,
|
980 |
+
"step": 9048
|
981 |
+
},
|
982 |
+
{
|
983 |
+
"best_epoch": 40,
|
984 |
+
"best_eval_accuracy": 0.7364620938628159,
|
985 |
+
"epoch": 58.0,
|
986 |
+
"step": 9048
|
987 |
+
},
|
988 |
+
{
|
989 |
+
"epoch": 59.0,
|
990 |
+
"eval_accuracy": 0.7148014440433214,
|
991 |
+
"eval_loss": 0.16792170703411102,
|
992 |
+
"eval_runtime": 4.3544,
|
993 |
+
"eval_samples_per_second": 63.614,
|
994 |
+
"eval_steps_per_second": 8.038,
|
995 |
+
"step": 9204
|
996 |
+
},
|
997 |
+
{
|
998 |
+
"best_epoch": 40,
|
999 |
+
"best_eval_accuracy": 0.7364620938628159,
|
1000 |
+
"epoch": 59.0,
|
1001 |
+
"step": 9204
|
1002 |
+
},
|
1003 |
+
{
|
1004 |
+
"epoch": 60.0,
|
1005 |
+
"eval_accuracy": 0.7148014440433214,
|
1006 |
+
"eval_loss": 0.16792602837085724,
|
1007 |
+
"eval_runtime": 4.3537,
|
1008 |
+
"eval_samples_per_second": 63.624,
|
1009 |
+
"eval_steps_per_second": 8.039,
|
1010 |
+
"step": 9360
|
1011 |
+
},
|
1012 |
+
{
|
1013 |
+
"best_epoch": 40,
|
1014 |
+
"best_eval_accuracy": 0.7364620938628159,
|
1015 |
+
"epoch": 60.0,
|
1016 |
+
"step": 9360
|
1017 |
+
},
|
1018 |
+
{
|
1019 |
+
"epoch": 60.0,
|
1020 |
+
"step": 9360,
|
1021 |
+
"total_flos": 6.96152728406016e+16,
|
1022 |
+
"train_loss": 0.18218991573040302,
|
1023 |
+
"train_runtime": 3746.5818,
|
1024 |
+
"train_samples_per_second": 39.876,
|
1025 |
+
"train_steps_per_second": 2.498
|
1026 |
+
}
|
1027 |
+
],
|
1028 |
+
"max_steps": 9360,
|
1029 |
+
"num_train_epochs": 60,
|
1030 |
+
"total_flos": 6.96152728406016e+16,
|
1031 |
+
"trial_name": null,
|
1032 |
+
"trial_params": null
|
1033 |
+
}
|