End of training
Browse files- all_results.json +7 -0
- best_results.json +5 -0
- train_results.json +7 -0
- trainer_state.json +1249 -0
all_results.json
ADDED
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 80.0,
|
3 |
+
"train_loss": 0.4624418716430664,
|
4 |
+
"train_runtime": 1699.8447,
|
5 |
+
"train_samples_per_second": 18.825,
|
6 |
+
"train_steps_per_second": 1.177
|
7 |
+
}
|
best_results.json
ADDED
@@ -0,0 +1,5 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_epoch": 16,
|
3 |
+
"best_eval_accuracy": 0.69,
|
4 |
+
"epoch": 80.0
|
5 |
+
}
|
train_results.json
ADDED
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 80.0,
|
3 |
+
"train_loss": 0.4624418716430664,
|
4 |
+
"train_runtime": 1699.8447,
|
5 |
+
"train_samples_per_second": 18.825,
|
6 |
+
"train_steps_per_second": 1.177
|
7 |
+
}
|
trainer_state.json
ADDED
@@ -0,0 +1,1249 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 80.0,
|
5 |
+
"global_step": 2000,
|
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": 0.64,
|
13 |
+
"eval_loss": 0.29521188139915466,
|
14 |
+
"eval_runtime": 2.9613,
|
15 |
+
"eval_samples_per_second": 33.769,
|
16 |
+
"eval_steps_per_second": 4.39,
|
17 |
+
"step": 25
|
18 |
+
},
|
19 |
+
{
|
20 |
+
"best_epoch": 0,
|
21 |
+
"best_eval_accuracy": 0.64,
|
22 |
+
"epoch": 1.0,
|
23 |
+
"step": 25
|
24 |
+
},
|
25 |
+
{
|
26 |
+
"epoch": 2.0,
|
27 |
+
"eval_accuracy": 0.57,
|
28 |
+
"eval_loss": 0.28949299454689026,
|
29 |
+
"eval_runtime": 3.029,
|
30 |
+
"eval_samples_per_second": 33.015,
|
31 |
+
"eval_steps_per_second": 4.292,
|
32 |
+
"step": 50
|
33 |
+
},
|
34 |
+
{
|
35 |
+
"best_epoch": 0,
|
36 |
+
"best_eval_accuracy": 0.64,
|
37 |
+
"epoch": 2.0,
|
38 |
+
"step": 50
|
39 |
+
},
|
40 |
+
{
|
41 |
+
"epoch": 3.0,
|
42 |
+
"eval_accuracy": 0.61,
|
43 |
+
"eval_loss": 0.29219087958335876,
|
44 |
+
"eval_runtime": 3.0806,
|
45 |
+
"eval_samples_per_second": 32.461,
|
46 |
+
"eval_steps_per_second": 4.22,
|
47 |
+
"step": 75
|
48 |
+
},
|
49 |
+
{
|
50 |
+
"best_epoch": 0,
|
51 |
+
"best_eval_accuracy": 0.64,
|
52 |
+
"epoch": 3.0,
|
53 |
+
"step": 75
|
54 |
+
},
|
55 |
+
{
|
56 |
+
"epoch": 4.0,
|
57 |
+
"eval_accuracy": 0.64,
|
58 |
+
"eval_loss": 0.29383212327957153,
|
59 |
+
"eval_runtime": 3.1144,
|
60 |
+
"eval_samples_per_second": 32.109,
|
61 |
+
"eval_steps_per_second": 4.174,
|
62 |
+
"step": 100
|
63 |
+
},
|
64 |
+
{
|
65 |
+
"best_epoch": 0,
|
66 |
+
"best_eval_accuracy": 0.64,
|
67 |
+
"epoch": 4.0,
|
68 |
+
"step": 100
|
69 |
+
},
|
70 |
+
{
|
71 |
+
"epoch": 5.0,
|
72 |
+
"eval_accuracy": 0.63,
|
73 |
+
"eval_loss": 0.28848135471343994,
|
74 |
+
"eval_runtime": 3.1511,
|
75 |
+
"eval_samples_per_second": 31.735,
|
76 |
+
"eval_steps_per_second": 4.126,
|
77 |
+
"step": 125
|
78 |
+
},
|
79 |
+
{
|
80 |
+
"best_epoch": 0,
|
81 |
+
"best_eval_accuracy": 0.64,
|
82 |
+
"epoch": 5.0,
|
83 |
+
"step": 125
|
84 |
+
},
|
85 |
+
{
|
86 |
+
"epoch": 6.0,
|
87 |
+
"eval_accuracy": 0.48,
|
88 |
+
"eval_loss": 0.2944973409175873,
|
89 |
+
"eval_runtime": 3.1574,
|
90 |
+
"eval_samples_per_second": 31.672,
|
91 |
+
"eval_steps_per_second": 4.117,
|
92 |
+
"step": 150
|
93 |
+
},
|
94 |
+
{
|
95 |
+
"best_epoch": 0,
|
96 |
+
"best_eval_accuracy": 0.64,
|
97 |
+
"epoch": 6.0,
|
98 |
+
"step": 150
|
99 |
+
},
|
100 |
+
{
|
101 |
+
"epoch": 7.0,
|
102 |
+
"eval_accuracy": 0.67,
|
103 |
+
"eval_loss": 0.2860355079174042,
|
104 |
+
"eval_runtime": 3.1678,
|
105 |
+
"eval_samples_per_second": 31.568,
|
106 |
+
"eval_steps_per_second": 4.104,
|
107 |
+
"step": 175
|
108 |
+
},
|
109 |
+
{
|
110 |
+
"best_epoch": 6,
|
111 |
+
"best_eval_accuracy": 0.67,
|
112 |
+
"epoch": 7.0,
|
113 |
+
"step": 175
|
114 |
+
},
|
115 |
+
{
|
116 |
+
"epoch": 8.0,
|
117 |
+
"eval_accuracy": 0.66,
|
118 |
+
"eval_loss": 0.2887631356716156,
|
119 |
+
"eval_runtime": 3.1726,
|
120 |
+
"eval_samples_per_second": 31.52,
|
121 |
+
"eval_steps_per_second": 4.098,
|
122 |
+
"step": 200
|
123 |
+
},
|
124 |
+
{
|
125 |
+
"best_epoch": 6,
|
126 |
+
"best_eval_accuracy": 0.67,
|
127 |
+
"epoch": 8.0,
|
128 |
+
"step": 200
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 9.0,
|
132 |
+
"eval_accuracy": 0.51,
|
133 |
+
"eval_loss": 0.28938525915145874,
|
134 |
+
"eval_runtime": 3.1696,
|
135 |
+
"eval_samples_per_second": 31.55,
|
136 |
+
"eval_steps_per_second": 4.101,
|
137 |
+
"step": 225
|
138 |
+
},
|
139 |
+
{
|
140 |
+
"best_epoch": 6,
|
141 |
+
"best_eval_accuracy": 0.67,
|
142 |
+
"epoch": 9.0,
|
143 |
+
"step": 225
|
144 |
+
},
|
145 |
+
{
|
146 |
+
"epoch": 10.0,
|
147 |
+
"eval_accuracy": 0.56,
|
148 |
+
"eval_loss": 0.29034924507141113,
|
149 |
+
"eval_runtime": 3.1725,
|
150 |
+
"eval_samples_per_second": 31.521,
|
151 |
+
"eval_steps_per_second": 4.098,
|
152 |
+
"step": 250
|
153 |
+
},
|
154 |
+
{
|
155 |
+
"best_epoch": 6,
|
156 |
+
"best_eval_accuracy": 0.67,
|
157 |
+
"epoch": 10.0,
|
158 |
+
"step": 250
|
159 |
+
},
|
160 |
+
{
|
161 |
+
"epoch": 11.0,
|
162 |
+
"eval_accuracy": 0.66,
|
163 |
+
"eval_loss": 0.286794513463974,
|
164 |
+
"eval_runtime": 3.1738,
|
165 |
+
"eval_samples_per_second": 31.508,
|
166 |
+
"eval_steps_per_second": 4.096,
|
167 |
+
"step": 275
|
168 |
+
},
|
169 |
+
{
|
170 |
+
"best_epoch": 6,
|
171 |
+
"best_eval_accuracy": 0.67,
|
172 |
+
"epoch": 11.0,
|
173 |
+
"step": 275
|
174 |
+
},
|
175 |
+
{
|
176 |
+
"epoch": 12.0,
|
177 |
+
"eval_accuracy": 0.66,
|
178 |
+
"eval_loss": 0.287968248128891,
|
179 |
+
"eval_runtime": 3.1761,
|
180 |
+
"eval_samples_per_second": 31.485,
|
181 |
+
"eval_steps_per_second": 4.093,
|
182 |
+
"step": 300
|
183 |
+
},
|
184 |
+
{
|
185 |
+
"best_epoch": 6,
|
186 |
+
"best_eval_accuracy": 0.67,
|
187 |
+
"epoch": 12.0,
|
188 |
+
"step": 300
|
189 |
+
},
|
190 |
+
{
|
191 |
+
"epoch": 13.0,
|
192 |
+
"eval_accuracy": 0.54,
|
193 |
+
"eval_loss": 0.2946690618991852,
|
194 |
+
"eval_runtime": 3.1756,
|
195 |
+
"eval_samples_per_second": 31.49,
|
196 |
+
"eval_steps_per_second": 4.094,
|
197 |
+
"step": 325
|
198 |
+
},
|
199 |
+
{
|
200 |
+
"best_epoch": 6,
|
201 |
+
"best_eval_accuracy": 0.67,
|
202 |
+
"epoch": 13.0,
|
203 |
+
"step": 325
|
204 |
+
},
|
205 |
+
{
|
206 |
+
"epoch": 14.0,
|
207 |
+
"eval_accuracy": 0.64,
|
208 |
+
"eval_loss": 0.29572564363479614,
|
209 |
+
"eval_runtime": 3.1786,
|
210 |
+
"eval_samples_per_second": 31.46,
|
211 |
+
"eval_steps_per_second": 4.09,
|
212 |
+
"step": 350
|
213 |
+
},
|
214 |
+
{
|
215 |
+
"best_epoch": 6,
|
216 |
+
"best_eval_accuracy": 0.67,
|
217 |
+
"epoch": 14.0,
|
218 |
+
"step": 350
|
219 |
+
},
|
220 |
+
{
|
221 |
+
"epoch": 15.0,
|
222 |
+
"eval_accuracy": 0.66,
|
223 |
+
"eval_loss": 0.2876897156238556,
|
224 |
+
"eval_runtime": 3.1783,
|
225 |
+
"eval_samples_per_second": 31.463,
|
226 |
+
"eval_steps_per_second": 4.09,
|
227 |
+
"step": 375
|
228 |
+
},
|
229 |
+
{
|
230 |
+
"best_epoch": 6,
|
231 |
+
"best_eval_accuracy": 0.67,
|
232 |
+
"epoch": 15.0,
|
233 |
+
"step": 375
|
234 |
+
},
|
235 |
+
{
|
236 |
+
"epoch": 16.0,
|
237 |
+
"eval_accuracy": 0.68,
|
238 |
+
"eval_loss": 0.28647202253341675,
|
239 |
+
"eval_runtime": 3.1833,
|
240 |
+
"eval_samples_per_second": 31.414,
|
241 |
+
"eval_steps_per_second": 4.084,
|
242 |
+
"step": 400
|
243 |
+
},
|
244 |
+
{
|
245 |
+
"best_epoch": 15,
|
246 |
+
"best_eval_accuracy": 0.68,
|
247 |
+
"epoch": 16.0,
|
248 |
+
"step": 400
|
249 |
+
},
|
250 |
+
{
|
251 |
+
"epoch": 17.0,
|
252 |
+
"eval_accuracy": 0.69,
|
253 |
+
"eval_loss": 0.28502991795539856,
|
254 |
+
"eval_runtime": 3.1839,
|
255 |
+
"eval_samples_per_second": 31.408,
|
256 |
+
"eval_steps_per_second": 4.083,
|
257 |
+
"step": 425
|
258 |
+
},
|
259 |
+
{
|
260 |
+
"best_epoch": 16,
|
261 |
+
"best_eval_accuracy": 0.69,
|
262 |
+
"epoch": 17.0,
|
263 |
+
"step": 425
|
264 |
+
},
|
265 |
+
{
|
266 |
+
"epoch": 18.0,
|
267 |
+
"eval_accuracy": 0.66,
|
268 |
+
"eval_loss": 0.28460079431533813,
|
269 |
+
"eval_runtime": 3.1831,
|
270 |
+
"eval_samples_per_second": 31.416,
|
271 |
+
"eval_steps_per_second": 4.084,
|
272 |
+
"step": 450
|
273 |
+
},
|
274 |
+
{
|
275 |
+
"best_epoch": 16,
|
276 |
+
"best_eval_accuracy": 0.69,
|
277 |
+
"epoch": 18.0,
|
278 |
+
"step": 450
|
279 |
+
},
|
280 |
+
{
|
281 |
+
"epoch": 19.0,
|
282 |
+
"eval_accuracy": 0.59,
|
283 |
+
"eval_loss": 0.29105326533317566,
|
284 |
+
"eval_runtime": 3.1856,
|
285 |
+
"eval_samples_per_second": 31.391,
|
286 |
+
"eval_steps_per_second": 4.081,
|
287 |
+
"step": 475
|
288 |
+
},
|
289 |
+
{
|
290 |
+
"best_epoch": 16,
|
291 |
+
"best_eval_accuracy": 0.69,
|
292 |
+
"epoch": 19.0,
|
293 |
+
"step": 475
|
294 |
+
},
|
295 |
+
{
|
296 |
+
"epoch": 20.0,
|
297 |
+
"learning_rate": 0.00075,
|
298 |
+
"loss": 0.4684,
|
299 |
+
"step": 500
|
300 |
+
},
|
301 |
+
{
|
302 |
+
"epoch": 20.0,
|
303 |
+
"eval_accuracy": 0.64,
|
304 |
+
"eval_loss": 0.2960752844810486,
|
305 |
+
"eval_runtime": 3.1862,
|
306 |
+
"eval_samples_per_second": 31.386,
|
307 |
+
"eval_steps_per_second": 4.08,
|
308 |
+
"step": 500
|
309 |
+
},
|
310 |
+
{
|
311 |
+
"best_epoch": 16,
|
312 |
+
"best_eval_accuracy": 0.69,
|
313 |
+
"epoch": 20.0,
|
314 |
+
"step": 500
|
315 |
+
},
|
316 |
+
{
|
317 |
+
"epoch": 21.0,
|
318 |
+
"eval_accuracy": 0.63,
|
319 |
+
"eval_loss": 0.28715649247169495,
|
320 |
+
"eval_runtime": 3.1889,
|
321 |
+
"eval_samples_per_second": 31.359,
|
322 |
+
"eval_steps_per_second": 4.077,
|
323 |
+
"step": 525
|
324 |
+
},
|
325 |
+
{
|
326 |
+
"best_epoch": 16,
|
327 |
+
"best_eval_accuracy": 0.69,
|
328 |
+
"epoch": 21.0,
|
329 |
+
"step": 525
|
330 |
+
},
|
331 |
+
{
|
332 |
+
"epoch": 22.0,
|
333 |
+
"eval_accuracy": 0.64,
|
334 |
+
"eval_loss": 0.2879977822303772,
|
335 |
+
"eval_runtime": 3.1931,
|
336 |
+
"eval_samples_per_second": 31.317,
|
337 |
+
"eval_steps_per_second": 4.071,
|
338 |
+
"step": 550
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"best_epoch": 16,
|
342 |
+
"best_eval_accuracy": 0.69,
|
343 |
+
"epoch": 22.0,
|
344 |
+
"step": 550
|
345 |
+
},
|
346 |
+
{
|
347 |
+
"epoch": 23.0,
|
348 |
+
"eval_accuracy": 0.51,
|
349 |
+
"eval_loss": 0.2950732111930847,
|
350 |
+
"eval_runtime": 3.1897,
|
351 |
+
"eval_samples_per_second": 31.351,
|
352 |
+
"eval_steps_per_second": 4.076,
|
353 |
+
"step": 575
|
354 |
+
},
|
355 |
+
{
|
356 |
+
"best_epoch": 16,
|
357 |
+
"best_eval_accuracy": 0.69,
|
358 |
+
"epoch": 23.0,
|
359 |
+
"step": 575
|
360 |
+
},
|
361 |
+
{
|
362 |
+
"epoch": 24.0,
|
363 |
+
"eval_accuracy": 0.64,
|
364 |
+
"eval_loss": 0.2896903157234192,
|
365 |
+
"eval_runtime": 3.1901,
|
366 |
+
"eval_samples_per_second": 31.347,
|
367 |
+
"eval_steps_per_second": 4.075,
|
368 |
+
"step": 600
|
369 |
+
},
|
370 |
+
{
|
371 |
+
"best_epoch": 16,
|
372 |
+
"best_eval_accuracy": 0.69,
|
373 |
+
"epoch": 24.0,
|
374 |
+
"step": 600
|
375 |
+
},
|
376 |
+
{
|
377 |
+
"epoch": 25.0,
|
378 |
+
"eval_accuracy": 0.64,
|
379 |
+
"eval_loss": 0.2884129285812378,
|
380 |
+
"eval_runtime": 3.1902,
|
381 |
+
"eval_samples_per_second": 31.346,
|
382 |
+
"eval_steps_per_second": 4.075,
|
383 |
+
"step": 625
|
384 |
+
},
|
385 |
+
{
|
386 |
+
"best_epoch": 16,
|
387 |
+
"best_eval_accuracy": 0.69,
|
388 |
+
"epoch": 25.0,
|
389 |
+
"step": 625
|
390 |
+
},
|
391 |
+
{
|
392 |
+
"epoch": 26.0,
|
393 |
+
"eval_accuracy": 0.64,
|
394 |
+
"eval_loss": 0.2894528806209564,
|
395 |
+
"eval_runtime": 3.1904,
|
396 |
+
"eval_samples_per_second": 31.344,
|
397 |
+
"eval_steps_per_second": 4.075,
|
398 |
+
"step": 650
|
399 |
+
},
|
400 |
+
{
|
401 |
+
"best_epoch": 16,
|
402 |
+
"best_eval_accuracy": 0.69,
|
403 |
+
"epoch": 26.0,
|
404 |
+
"step": 650
|
405 |
+
},
|
406 |
+
{
|
407 |
+
"epoch": 27.0,
|
408 |
+
"eval_accuracy": 0.61,
|
409 |
+
"eval_loss": 0.2871526777744293,
|
410 |
+
"eval_runtime": 3.1931,
|
411 |
+
"eval_samples_per_second": 31.317,
|
412 |
+
"eval_steps_per_second": 4.071,
|
413 |
+
"step": 675
|
414 |
+
},
|
415 |
+
{
|
416 |
+
"best_epoch": 16,
|
417 |
+
"best_eval_accuracy": 0.69,
|
418 |
+
"epoch": 27.0,
|
419 |
+
"step": 675
|
420 |
+
},
|
421 |
+
{
|
422 |
+
"epoch": 28.0,
|
423 |
+
"eval_accuracy": 0.64,
|
424 |
+
"eval_loss": 0.28898343443870544,
|
425 |
+
"eval_runtime": 3.1862,
|
426 |
+
"eval_samples_per_second": 31.385,
|
427 |
+
"eval_steps_per_second": 4.08,
|
428 |
+
"step": 700
|
429 |
+
},
|
430 |
+
{
|
431 |
+
"best_epoch": 16,
|
432 |
+
"best_eval_accuracy": 0.69,
|
433 |
+
"epoch": 28.0,
|
434 |
+
"step": 700
|
435 |
+
},
|
436 |
+
{
|
437 |
+
"epoch": 29.0,
|
438 |
+
"eval_accuracy": 0.66,
|
439 |
+
"eval_loss": 0.2887178659439087,
|
440 |
+
"eval_runtime": 3.1838,
|
441 |
+
"eval_samples_per_second": 31.409,
|
442 |
+
"eval_steps_per_second": 4.083,
|
443 |
+
"step": 725
|
444 |
+
},
|
445 |
+
{
|
446 |
+
"best_epoch": 16,
|
447 |
+
"best_eval_accuracy": 0.69,
|
448 |
+
"epoch": 29.0,
|
449 |
+
"step": 725
|
450 |
+
},
|
451 |
+
{
|
452 |
+
"epoch": 30.0,
|
453 |
+
"eval_accuracy": 0.63,
|
454 |
+
"eval_loss": 0.2886087894439697,
|
455 |
+
"eval_runtime": 3.1839,
|
456 |
+
"eval_samples_per_second": 31.408,
|
457 |
+
"eval_steps_per_second": 4.083,
|
458 |
+
"step": 750
|
459 |
+
},
|
460 |
+
{
|
461 |
+
"best_epoch": 16,
|
462 |
+
"best_eval_accuracy": 0.69,
|
463 |
+
"epoch": 30.0,
|
464 |
+
"step": 750
|
465 |
+
},
|
466 |
+
{
|
467 |
+
"epoch": 31.0,
|
468 |
+
"eval_accuracy": 0.6,
|
469 |
+
"eval_loss": 0.2875303328037262,
|
470 |
+
"eval_runtime": 3.1843,
|
471 |
+
"eval_samples_per_second": 31.404,
|
472 |
+
"eval_steps_per_second": 4.082,
|
473 |
+
"step": 775
|
474 |
+
},
|
475 |
+
{
|
476 |
+
"best_epoch": 16,
|
477 |
+
"best_eval_accuracy": 0.69,
|
478 |
+
"epoch": 31.0,
|
479 |
+
"step": 775
|
480 |
+
},
|
481 |
+
{
|
482 |
+
"epoch": 32.0,
|
483 |
+
"eval_accuracy": 0.65,
|
484 |
+
"eval_loss": 0.2882201075553894,
|
485 |
+
"eval_runtime": 3.1843,
|
486 |
+
"eval_samples_per_second": 31.404,
|
487 |
+
"eval_steps_per_second": 4.083,
|
488 |
+
"step": 800
|
489 |
+
},
|
490 |
+
{
|
491 |
+
"best_epoch": 16,
|
492 |
+
"best_eval_accuracy": 0.69,
|
493 |
+
"epoch": 32.0,
|
494 |
+
"step": 800
|
495 |
+
},
|
496 |
+
{
|
497 |
+
"epoch": 33.0,
|
498 |
+
"eval_accuracy": 0.58,
|
499 |
+
"eval_loss": 0.2886188328266144,
|
500 |
+
"eval_runtime": 3.1833,
|
501 |
+
"eval_samples_per_second": 31.414,
|
502 |
+
"eval_steps_per_second": 4.084,
|
503 |
+
"step": 825
|
504 |
+
},
|
505 |
+
{
|
506 |
+
"best_epoch": 16,
|
507 |
+
"best_eval_accuracy": 0.69,
|
508 |
+
"epoch": 33.0,
|
509 |
+
"step": 825
|
510 |
+
},
|
511 |
+
{
|
512 |
+
"epoch": 34.0,
|
513 |
+
"eval_accuracy": 0.64,
|
514 |
+
"eval_loss": 0.2970466911792755,
|
515 |
+
"eval_runtime": 3.1842,
|
516 |
+
"eval_samples_per_second": 31.405,
|
517 |
+
"eval_steps_per_second": 4.083,
|
518 |
+
"step": 850
|
519 |
+
},
|
520 |
+
{
|
521 |
+
"best_epoch": 16,
|
522 |
+
"best_eval_accuracy": 0.69,
|
523 |
+
"epoch": 34.0,
|
524 |
+
"step": 850
|
525 |
+
},
|
526 |
+
{
|
527 |
+
"epoch": 35.0,
|
528 |
+
"eval_accuracy": 0.59,
|
529 |
+
"eval_loss": 0.287495493888855,
|
530 |
+
"eval_runtime": 3.1852,
|
531 |
+
"eval_samples_per_second": 31.396,
|
532 |
+
"eval_steps_per_second": 4.081,
|
533 |
+
"step": 875
|
534 |
+
},
|
535 |
+
{
|
536 |
+
"best_epoch": 16,
|
537 |
+
"best_eval_accuracy": 0.69,
|
538 |
+
"epoch": 35.0,
|
539 |
+
"step": 875
|
540 |
+
},
|
541 |
+
{
|
542 |
+
"epoch": 36.0,
|
543 |
+
"eval_accuracy": 0.63,
|
544 |
+
"eval_loss": 0.2887771725654602,
|
545 |
+
"eval_runtime": 3.1879,
|
546 |
+
"eval_samples_per_second": 31.368,
|
547 |
+
"eval_steps_per_second": 4.078,
|
548 |
+
"step": 900
|
549 |
+
},
|
550 |
+
{
|
551 |
+
"best_epoch": 16,
|
552 |
+
"best_eval_accuracy": 0.69,
|
553 |
+
"epoch": 36.0,
|
554 |
+
"step": 900
|
555 |
+
},
|
556 |
+
{
|
557 |
+
"epoch": 37.0,
|
558 |
+
"eval_accuracy": 0.63,
|
559 |
+
"eval_loss": 0.2868404686450958,
|
560 |
+
"eval_runtime": 3.1879,
|
561 |
+
"eval_samples_per_second": 31.369,
|
562 |
+
"eval_steps_per_second": 4.078,
|
563 |
+
"step": 925
|
564 |
+
},
|
565 |
+
{
|
566 |
+
"best_epoch": 16,
|
567 |
+
"best_eval_accuracy": 0.69,
|
568 |
+
"epoch": 37.0,
|
569 |
+
"step": 925
|
570 |
+
},
|
571 |
+
{
|
572 |
+
"epoch": 38.0,
|
573 |
+
"eval_accuracy": 0.64,
|
574 |
+
"eval_loss": 0.28633153438568115,
|
575 |
+
"eval_runtime": 3.1886,
|
576 |
+
"eval_samples_per_second": 31.361,
|
577 |
+
"eval_steps_per_second": 4.077,
|
578 |
+
"step": 950
|
579 |
+
},
|
580 |
+
{
|
581 |
+
"best_epoch": 16,
|
582 |
+
"best_eval_accuracy": 0.69,
|
583 |
+
"epoch": 38.0,
|
584 |
+
"step": 950
|
585 |
+
},
|
586 |
+
{
|
587 |
+
"epoch": 39.0,
|
588 |
+
"eval_accuracy": 0.63,
|
589 |
+
"eval_loss": 0.2910645008087158,
|
590 |
+
"eval_runtime": 3.1871,
|
591 |
+
"eval_samples_per_second": 31.376,
|
592 |
+
"eval_steps_per_second": 4.079,
|
593 |
+
"step": 975
|
594 |
+
},
|
595 |
+
{
|
596 |
+
"best_epoch": 16,
|
597 |
+
"best_eval_accuracy": 0.69,
|
598 |
+
"epoch": 39.0,
|
599 |
+
"step": 975
|
600 |
+
},
|
601 |
+
{
|
602 |
+
"epoch": 40.0,
|
603 |
+
"learning_rate": 0.0005,
|
604 |
+
"loss": 0.4634,
|
605 |
+
"step": 1000
|
606 |
+
},
|
607 |
+
{
|
608 |
+
"epoch": 40.0,
|
609 |
+
"eval_accuracy": 0.63,
|
610 |
+
"eval_loss": 0.2866987884044647,
|
611 |
+
"eval_runtime": 3.1867,
|
612 |
+
"eval_samples_per_second": 31.38,
|
613 |
+
"eval_steps_per_second": 4.079,
|
614 |
+
"step": 1000
|
615 |
+
},
|
616 |
+
{
|
617 |
+
"best_epoch": 16,
|
618 |
+
"best_eval_accuracy": 0.69,
|
619 |
+
"epoch": 40.0,
|
620 |
+
"step": 1000
|
621 |
+
},
|
622 |
+
{
|
623 |
+
"epoch": 41.0,
|
624 |
+
"eval_accuracy": 0.54,
|
625 |
+
"eval_loss": 0.29356110095977783,
|
626 |
+
"eval_runtime": 3.1874,
|
627 |
+
"eval_samples_per_second": 31.374,
|
628 |
+
"eval_steps_per_second": 4.079,
|
629 |
+
"step": 1025
|
630 |
+
},
|
631 |
+
{
|
632 |
+
"best_epoch": 16,
|
633 |
+
"best_eval_accuracy": 0.69,
|
634 |
+
"epoch": 41.0,
|
635 |
+
"step": 1025
|
636 |
+
},
|
637 |
+
{
|
638 |
+
"epoch": 42.0,
|
639 |
+
"eval_accuracy": 0.6,
|
640 |
+
"eval_loss": 0.29651960730552673,
|
641 |
+
"eval_runtime": 3.1917,
|
642 |
+
"eval_samples_per_second": 31.331,
|
643 |
+
"eval_steps_per_second": 4.073,
|
644 |
+
"step": 1050
|
645 |
+
},
|
646 |
+
{
|
647 |
+
"best_epoch": 16,
|
648 |
+
"best_eval_accuracy": 0.69,
|
649 |
+
"epoch": 42.0,
|
650 |
+
"step": 1050
|
651 |
+
},
|
652 |
+
{
|
653 |
+
"epoch": 43.0,
|
654 |
+
"eval_accuracy": 0.62,
|
655 |
+
"eval_loss": 0.2872467339038849,
|
656 |
+
"eval_runtime": 3.1905,
|
657 |
+
"eval_samples_per_second": 31.344,
|
658 |
+
"eval_steps_per_second": 4.075,
|
659 |
+
"step": 1075
|
660 |
+
},
|
661 |
+
{
|
662 |
+
"best_epoch": 16,
|
663 |
+
"best_eval_accuracy": 0.69,
|
664 |
+
"epoch": 43.0,
|
665 |
+
"step": 1075
|
666 |
+
},
|
667 |
+
{
|
668 |
+
"epoch": 44.0,
|
669 |
+
"eval_accuracy": 0.65,
|
670 |
+
"eval_loss": 0.2862147092819214,
|
671 |
+
"eval_runtime": 3.1876,
|
672 |
+
"eval_samples_per_second": 31.372,
|
673 |
+
"eval_steps_per_second": 4.078,
|
674 |
+
"step": 1100
|
675 |
+
},
|
676 |
+
{
|
677 |
+
"best_epoch": 16,
|
678 |
+
"best_eval_accuracy": 0.69,
|
679 |
+
"epoch": 44.0,
|
680 |
+
"step": 1100
|
681 |
+
},
|
682 |
+
{
|
683 |
+
"epoch": 45.0,
|
684 |
+
"eval_accuracy": 0.65,
|
685 |
+
"eval_loss": 0.28714072704315186,
|
686 |
+
"eval_runtime": 3.1836,
|
687 |
+
"eval_samples_per_second": 31.411,
|
688 |
+
"eval_steps_per_second": 4.083,
|
689 |
+
"step": 1125
|
690 |
+
},
|
691 |
+
{
|
692 |
+
"best_epoch": 16,
|
693 |
+
"best_eval_accuracy": 0.69,
|
694 |
+
"epoch": 45.0,
|
695 |
+
"step": 1125
|
696 |
+
},
|
697 |
+
{
|
698 |
+
"epoch": 46.0,
|
699 |
+
"eval_accuracy": 0.63,
|
700 |
+
"eval_loss": 0.2913626432418823,
|
701 |
+
"eval_runtime": 3.1841,
|
702 |
+
"eval_samples_per_second": 31.406,
|
703 |
+
"eval_steps_per_second": 4.083,
|
704 |
+
"step": 1150
|
705 |
+
},
|
706 |
+
{
|
707 |
+
"best_epoch": 16,
|
708 |
+
"best_eval_accuracy": 0.69,
|
709 |
+
"epoch": 46.0,
|
710 |
+
"step": 1150
|
711 |
+
},
|
712 |
+
{
|
713 |
+
"epoch": 47.0,
|
714 |
+
"eval_accuracy": 0.64,
|
715 |
+
"eval_loss": 0.29249587655067444,
|
716 |
+
"eval_runtime": 3.1869,
|
717 |
+
"eval_samples_per_second": 31.378,
|
718 |
+
"eval_steps_per_second": 4.079,
|
719 |
+
"step": 1175
|
720 |
+
},
|
721 |
+
{
|
722 |
+
"best_epoch": 16,
|
723 |
+
"best_eval_accuracy": 0.69,
|
724 |
+
"epoch": 47.0,
|
725 |
+
"step": 1175
|
726 |
+
},
|
727 |
+
{
|
728 |
+
"epoch": 48.0,
|
729 |
+
"eval_accuracy": 0.64,
|
730 |
+
"eval_loss": 0.28830277919769287,
|
731 |
+
"eval_runtime": 3.1859,
|
732 |
+
"eval_samples_per_second": 31.388,
|
733 |
+
"eval_steps_per_second": 4.08,
|
734 |
+
"step": 1200
|
735 |
+
},
|
736 |
+
{
|
737 |
+
"best_epoch": 16,
|
738 |
+
"best_eval_accuracy": 0.69,
|
739 |
+
"epoch": 48.0,
|
740 |
+
"step": 1200
|
741 |
+
},
|
742 |
+
{
|
743 |
+
"epoch": 49.0,
|
744 |
+
"eval_accuracy": 0.65,
|
745 |
+
"eval_loss": 0.28956958651542664,
|
746 |
+
"eval_runtime": 3.1845,
|
747 |
+
"eval_samples_per_second": 31.402,
|
748 |
+
"eval_steps_per_second": 4.082,
|
749 |
+
"step": 1225
|
750 |
+
},
|
751 |
+
{
|
752 |
+
"best_epoch": 16,
|
753 |
+
"best_eval_accuracy": 0.69,
|
754 |
+
"epoch": 49.0,
|
755 |
+
"step": 1225
|
756 |
+
},
|
757 |
+
{
|
758 |
+
"epoch": 50.0,
|
759 |
+
"eval_accuracy": 0.64,
|
760 |
+
"eval_loss": 0.28662189841270447,
|
761 |
+
"eval_runtime": 3.1871,
|
762 |
+
"eval_samples_per_second": 31.377,
|
763 |
+
"eval_steps_per_second": 4.079,
|
764 |
+
"step": 1250
|
765 |
+
},
|
766 |
+
{
|
767 |
+
"best_epoch": 16,
|
768 |
+
"best_eval_accuracy": 0.69,
|
769 |
+
"epoch": 50.0,
|
770 |
+
"step": 1250
|
771 |
+
},
|
772 |
+
{
|
773 |
+
"epoch": 51.0,
|
774 |
+
"eval_accuracy": 0.64,
|
775 |
+
"eval_loss": 0.28566181659698486,
|
776 |
+
"eval_runtime": 3.1876,
|
777 |
+
"eval_samples_per_second": 31.371,
|
778 |
+
"eval_steps_per_second": 4.078,
|
779 |
+
"step": 1275
|
780 |
+
},
|
781 |
+
{
|
782 |
+
"best_epoch": 16,
|
783 |
+
"best_eval_accuracy": 0.69,
|
784 |
+
"epoch": 51.0,
|
785 |
+
"step": 1275
|
786 |
+
},
|
787 |
+
{
|
788 |
+
"epoch": 52.0,
|
789 |
+
"eval_accuracy": 0.64,
|
790 |
+
"eval_loss": 0.28922975063323975,
|
791 |
+
"eval_runtime": 3.186,
|
792 |
+
"eval_samples_per_second": 31.387,
|
793 |
+
"eval_steps_per_second": 4.08,
|
794 |
+
"step": 1300
|
795 |
+
},
|
796 |
+
{
|
797 |
+
"best_epoch": 16,
|
798 |
+
"best_eval_accuracy": 0.69,
|
799 |
+
"epoch": 52.0,
|
800 |
+
"step": 1300
|
801 |
+
},
|
802 |
+
{
|
803 |
+
"epoch": 53.0,
|
804 |
+
"eval_accuracy": 0.65,
|
805 |
+
"eval_loss": 0.2861008942127228,
|
806 |
+
"eval_runtime": 3.1847,
|
807 |
+
"eval_samples_per_second": 31.401,
|
808 |
+
"eval_steps_per_second": 4.082,
|
809 |
+
"step": 1325
|
810 |
+
},
|
811 |
+
{
|
812 |
+
"best_epoch": 16,
|
813 |
+
"best_eval_accuracy": 0.69,
|
814 |
+
"epoch": 53.0,
|
815 |
+
"step": 1325
|
816 |
+
},
|
817 |
+
{
|
818 |
+
"epoch": 54.0,
|
819 |
+
"eval_accuracy": 0.63,
|
820 |
+
"eval_loss": 0.2861438989639282,
|
821 |
+
"eval_runtime": 3.1893,
|
822 |
+
"eval_samples_per_second": 31.355,
|
823 |
+
"eval_steps_per_second": 4.076,
|
824 |
+
"step": 1350
|
825 |
+
},
|
826 |
+
{
|
827 |
+
"best_epoch": 16,
|
828 |
+
"best_eval_accuracy": 0.69,
|
829 |
+
"epoch": 54.0,
|
830 |
+
"step": 1350
|
831 |
+
},
|
832 |
+
{
|
833 |
+
"epoch": 55.0,
|
834 |
+
"eval_accuracy": 0.65,
|
835 |
+
"eval_loss": 0.2871514558792114,
|
836 |
+
"eval_runtime": 3.1842,
|
837 |
+
"eval_samples_per_second": 31.405,
|
838 |
+
"eval_steps_per_second": 4.083,
|
839 |
+
"step": 1375
|
840 |
+
},
|
841 |
+
{
|
842 |
+
"best_epoch": 16,
|
843 |
+
"best_eval_accuracy": 0.69,
|
844 |
+
"epoch": 55.0,
|
845 |
+
"step": 1375
|
846 |
+
},
|
847 |
+
{
|
848 |
+
"epoch": 56.0,
|
849 |
+
"eval_accuracy": 0.64,
|
850 |
+
"eval_loss": 0.2861255705356598,
|
851 |
+
"eval_runtime": 3.1867,
|
852 |
+
"eval_samples_per_second": 31.381,
|
853 |
+
"eval_steps_per_second": 4.08,
|
854 |
+
"step": 1400
|
855 |
+
},
|
856 |
+
{
|
857 |
+
"best_epoch": 16,
|
858 |
+
"best_eval_accuracy": 0.69,
|
859 |
+
"epoch": 56.0,
|
860 |
+
"step": 1400
|
861 |
+
},
|
862 |
+
{
|
863 |
+
"epoch": 57.0,
|
864 |
+
"eval_accuracy": 0.65,
|
865 |
+
"eval_loss": 0.2864649295806885,
|
866 |
+
"eval_runtime": 3.1885,
|
867 |
+
"eval_samples_per_second": 31.363,
|
868 |
+
"eval_steps_per_second": 4.077,
|
869 |
+
"step": 1425
|
870 |
+
},
|
871 |
+
{
|
872 |
+
"best_epoch": 16,
|
873 |
+
"best_eval_accuracy": 0.69,
|
874 |
+
"epoch": 57.0,
|
875 |
+
"step": 1425
|
876 |
+
},
|
877 |
+
{
|
878 |
+
"epoch": 58.0,
|
879 |
+
"eval_accuracy": 0.63,
|
880 |
+
"eval_loss": 0.2879580557346344,
|
881 |
+
"eval_runtime": 3.1877,
|
882 |
+
"eval_samples_per_second": 31.371,
|
883 |
+
"eval_steps_per_second": 4.078,
|
884 |
+
"step": 1450
|
885 |
+
},
|
886 |
+
{
|
887 |
+
"best_epoch": 16,
|
888 |
+
"best_eval_accuracy": 0.69,
|
889 |
+
"epoch": 58.0,
|
890 |
+
"step": 1450
|
891 |
+
},
|
892 |
+
{
|
893 |
+
"epoch": 59.0,
|
894 |
+
"eval_accuracy": 0.63,
|
895 |
+
"eval_loss": 0.2898498773574829,
|
896 |
+
"eval_runtime": 3.1921,
|
897 |
+
"eval_samples_per_second": 31.327,
|
898 |
+
"eval_steps_per_second": 4.073,
|
899 |
+
"step": 1475
|
900 |
+
},
|
901 |
+
{
|
902 |
+
"best_epoch": 16,
|
903 |
+
"best_eval_accuracy": 0.69,
|
904 |
+
"epoch": 59.0,
|
905 |
+
"step": 1475
|
906 |
+
},
|
907 |
+
{
|
908 |
+
"epoch": 60.0,
|
909 |
+
"learning_rate": 0.00025,
|
910 |
+
"loss": 0.4583,
|
911 |
+
"step": 1500
|
912 |
+
},
|
913 |
+
{
|
914 |
+
"epoch": 60.0,
|
915 |
+
"eval_accuracy": 0.63,
|
916 |
+
"eval_loss": 0.2899610102176666,
|
917 |
+
"eval_runtime": 3.1902,
|
918 |
+
"eval_samples_per_second": 31.346,
|
919 |
+
"eval_steps_per_second": 4.075,
|
920 |
+
"step": 1500
|
921 |
+
},
|
922 |
+
{
|
923 |
+
"best_epoch": 16,
|
924 |
+
"best_eval_accuracy": 0.69,
|
925 |
+
"epoch": 60.0,
|
926 |
+
"step": 1500
|
927 |
+
},
|
928 |
+
{
|
929 |
+
"epoch": 61.0,
|
930 |
+
"eval_accuracy": 0.64,
|
931 |
+
"eval_loss": 0.2895892858505249,
|
932 |
+
"eval_runtime": 3.1858,
|
933 |
+
"eval_samples_per_second": 31.389,
|
934 |
+
"eval_steps_per_second": 4.081,
|
935 |
+
"step": 1525
|
936 |
+
},
|
937 |
+
{
|
938 |
+
"best_epoch": 16,
|
939 |
+
"best_eval_accuracy": 0.69,
|
940 |
+
"epoch": 61.0,
|
941 |
+
"step": 1525
|
942 |
+
},
|
943 |
+
{
|
944 |
+
"epoch": 62.0,
|
945 |
+
"eval_accuracy": 0.63,
|
946 |
+
"eval_loss": 0.28861695528030396,
|
947 |
+
"eval_runtime": 3.1896,
|
948 |
+
"eval_samples_per_second": 31.352,
|
949 |
+
"eval_steps_per_second": 4.076,
|
950 |
+
"step": 1550
|
951 |
+
},
|
952 |
+
{
|
953 |
+
"best_epoch": 16,
|
954 |
+
"best_eval_accuracy": 0.69,
|
955 |
+
"epoch": 62.0,
|
956 |
+
"step": 1550
|
957 |
+
},
|
958 |
+
{
|
959 |
+
"epoch": 63.0,
|
960 |
+
"eval_accuracy": 0.63,
|
961 |
+
"eval_loss": 0.28878867626190186,
|
962 |
+
"eval_runtime": 3.1872,
|
963 |
+
"eval_samples_per_second": 31.375,
|
964 |
+
"eval_steps_per_second": 4.079,
|
965 |
+
"step": 1575
|
966 |
+
},
|
967 |
+
{
|
968 |
+
"best_epoch": 16,
|
969 |
+
"best_eval_accuracy": 0.69,
|
970 |
+
"epoch": 63.0,
|
971 |
+
"step": 1575
|
972 |
+
},
|
973 |
+
{
|
974 |
+
"epoch": 64.0,
|
975 |
+
"eval_accuracy": 0.64,
|
976 |
+
"eval_loss": 0.2890695333480835,
|
977 |
+
"eval_runtime": 3.1911,
|
978 |
+
"eval_samples_per_second": 31.337,
|
979 |
+
"eval_steps_per_second": 4.074,
|
980 |
+
"step": 1600
|
981 |
+
},
|
982 |
+
{
|
983 |
+
"best_epoch": 16,
|
984 |
+
"best_eval_accuracy": 0.69,
|
985 |
+
"epoch": 64.0,
|
986 |
+
"step": 1600
|
987 |
+
},
|
988 |
+
{
|
989 |
+
"epoch": 65.0,
|
990 |
+
"eval_accuracy": 0.63,
|
991 |
+
"eval_loss": 0.28738299012184143,
|
992 |
+
"eval_runtime": 3.1931,
|
993 |
+
"eval_samples_per_second": 31.317,
|
994 |
+
"eval_steps_per_second": 4.071,
|
995 |
+
"step": 1625
|
996 |
+
},
|
997 |
+
{
|
998 |
+
"best_epoch": 16,
|
999 |
+
"best_eval_accuracy": 0.69,
|
1000 |
+
"epoch": 65.0,
|
1001 |
+
"step": 1625
|
1002 |
+
},
|
1003 |
+
{
|
1004 |
+
"epoch": 66.0,
|
1005 |
+
"eval_accuracy": 0.62,
|
1006 |
+
"eval_loss": 0.28749388456344604,
|
1007 |
+
"eval_runtime": 3.1907,
|
1008 |
+
"eval_samples_per_second": 31.341,
|
1009 |
+
"eval_steps_per_second": 4.074,
|
1010 |
+
"step": 1650
|
1011 |
+
},
|
1012 |
+
{
|
1013 |
+
"best_epoch": 16,
|
1014 |
+
"best_eval_accuracy": 0.69,
|
1015 |
+
"epoch": 66.0,
|
1016 |
+
"step": 1650
|
1017 |
+
},
|
1018 |
+
{
|
1019 |
+
"epoch": 67.0,
|
1020 |
+
"eval_accuracy": 0.62,
|
1021 |
+
"eval_loss": 0.2881923019886017,
|
1022 |
+
"eval_runtime": 3.1995,
|
1023 |
+
"eval_samples_per_second": 31.255,
|
1024 |
+
"eval_steps_per_second": 4.063,
|
1025 |
+
"step": 1675
|
1026 |
+
},
|
1027 |
+
{
|
1028 |
+
"best_epoch": 16,
|
1029 |
+
"best_eval_accuracy": 0.69,
|
1030 |
+
"epoch": 67.0,
|
1031 |
+
"step": 1675
|
1032 |
+
},
|
1033 |
+
{
|
1034 |
+
"epoch": 68.0,
|
1035 |
+
"eval_accuracy": 0.62,
|
1036 |
+
"eval_loss": 0.2863309979438782,
|
1037 |
+
"eval_runtime": 3.1929,
|
1038 |
+
"eval_samples_per_second": 31.32,
|
1039 |
+
"eval_steps_per_second": 4.072,
|
1040 |
+
"step": 1700
|
1041 |
+
},
|
1042 |
+
{
|
1043 |
+
"best_epoch": 16,
|
1044 |
+
"best_eval_accuracy": 0.69,
|
1045 |
+
"epoch": 68.0,
|
1046 |
+
"step": 1700
|
1047 |
+
},
|
1048 |
+
{
|
1049 |
+
"epoch": 69.0,
|
1050 |
+
"eval_accuracy": 0.63,
|
1051 |
+
"eval_loss": 0.2866677939891815,
|
1052 |
+
"eval_runtime": 3.1936,
|
1053 |
+
"eval_samples_per_second": 31.313,
|
1054 |
+
"eval_steps_per_second": 4.071,
|
1055 |
+
"step": 1725
|
1056 |
+
},
|
1057 |
+
{
|
1058 |
+
"best_epoch": 16,
|
1059 |
+
"best_eval_accuracy": 0.69,
|
1060 |
+
"epoch": 69.0,
|
1061 |
+
"step": 1725
|
1062 |
+
},
|
1063 |
+
{
|
1064 |
+
"epoch": 70.0,
|
1065 |
+
"eval_accuracy": 0.64,
|
1066 |
+
"eval_loss": 0.28654342889785767,
|
1067 |
+
"eval_runtime": 3.1984,
|
1068 |
+
"eval_samples_per_second": 31.266,
|
1069 |
+
"eval_steps_per_second": 4.065,
|
1070 |
+
"step": 1750
|
1071 |
+
},
|
1072 |
+
{
|
1073 |
+
"best_epoch": 16,
|
1074 |
+
"best_eval_accuracy": 0.69,
|
1075 |
+
"epoch": 70.0,
|
1076 |
+
"step": 1750
|
1077 |
+
},
|
1078 |
+
{
|
1079 |
+
"epoch": 71.0,
|
1080 |
+
"eval_accuracy": 0.64,
|
1081 |
+
"eval_loss": 0.2863212823867798,
|
1082 |
+
"eval_runtime": 3.2026,
|
1083 |
+
"eval_samples_per_second": 31.225,
|
1084 |
+
"eval_steps_per_second": 4.059,
|
1085 |
+
"step": 1775
|
1086 |
+
},
|
1087 |
+
{
|
1088 |
+
"best_epoch": 16,
|
1089 |
+
"best_eval_accuracy": 0.69,
|
1090 |
+
"epoch": 71.0,
|
1091 |
+
"step": 1775
|
1092 |
+
},
|
1093 |
+
{
|
1094 |
+
"epoch": 72.0,
|
1095 |
+
"eval_accuracy": 0.64,
|
1096 |
+
"eval_loss": 0.2862430512905121,
|
1097 |
+
"eval_runtime": 3.2028,
|
1098 |
+
"eval_samples_per_second": 31.223,
|
1099 |
+
"eval_steps_per_second": 4.059,
|
1100 |
+
"step": 1800
|
1101 |
+
},
|
1102 |
+
{
|
1103 |
+
"best_epoch": 16,
|
1104 |
+
"best_eval_accuracy": 0.69,
|
1105 |
+
"epoch": 72.0,
|
1106 |
+
"step": 1800
|
1107 |
+
},
|
1108 |
+
{
|
1109 |
+
"epoch": 73.0,
|
1110 |
+
"eval_accuracy": 0.64,
|
1111 |
+
"eval_loss": 0.28635847568511963,
|
1112 |
+
"eval_runtime": 3.2023,
|
1113 |
+
"eval_samples_per_second": 31.227,
|
1114 |
+
"eval_steps_per_second": 4.06,
|
1115 |
+
"step": 1825
|
1116 |
+
},
|
1117 |
+
{
|
1118 |
+
"best_epoch": 16,
|
1119 |
+
"best_eval_accuracy": 0.69,
|
1120 |
+
"epoch": 73.0,
|
1121 |
+
"step": 1825
|
1122 |
+
},
|
1123 |
+
{
|
1124 |
+
"epoch": 74.0,
|
1125 |
+
"eval_accuracy": 0.64,
|
1126 |
+
"eval_loss": 0.28623470664024353,
|
1127 |
+
"eval_runtime": 3.2087,
|
1128 |
+
"eval_samples_per_second": 31.166,
|
1129 |
+
"eval_steps_per_second": 4.052,
|
1130 |
+
"step": 1850
|
1131 |
+
},
|
1132 |
+
{
|
1133 |
+
"best_epoch": 16,
|
1134 |
+
"best_eval_accuracy": 0.69,
|
1135 |
+
"epoch": 74.0,
|
1136 |
+
"step": 1850
|
1137 |
+
},
|
1138 |
+
{
|
1139 |
+
"epoch": 75.0,
|
1140 |
+
"eval_accuracy": 0.64,
|
1141 |
+
"eval_loss": 0.2866401672363281,
|
1142 |
+
"eval_runtime": 3.2061,
|
1143 |
+
"eval_samples_per_second": 31.19,
|
1144 |
+
"eval_steps_per_second": 4.055,
|
1145 |
+
"step": 1875
|
1146 |
+
},
|
1147 |
+
{
|
1148 |
+
"best_epoch": 16,
|
1149 |
+
"best_eval_accuracy": 0.69,
|
1150 |
+
"epoch": 75.0,
|
1151 |
+
"step": 1875
|
1152 |
+
},
|
1153 |
+
{
|
1154 |
+
"epoch": 76.0,
|
1155 |
+
"eval_accuracy": 0.63,
|
1156 |
+
"eval_loss": 0.28675001859664917,
|
1157 |
+
"eval_runtime": 3.2056,
|
1158 |
+
"eval_samples_per_second": 31.195,
|
1159 |
+
"eval_steps_per_second": 4.055,
|
1160 |
+
"step": 1900
|
1161 |
+
},
|
1162 |
+
{
|
1163 |
+
"best_epoch": 16,
|
1164 |
+
"best_eval_accuracy": 0.69,
|
1165 |
+
"epoch": 76.0,
|
1166 |
+
"step": 1900
|
1167 |
+
},
|
1168 |
+
{
|
1169 |
+
"epoch": 77.0,
|
1170 |
+
"eval_accuracy": 0.63,
|
1171 |
+
"eval_loss": 0.2866291403770447,
|
1172 |
+
"eval_runtime": 3.2051,
|
1173 |
+
"eval_samples_per_second": 31.201,
|
1174 |
+
"eval_steps_per_second": 4.056,
|
1175 |
+
"step": 1925
|
1176 |
+
},
|
1177 |
+
{
|
1178 |
+
"best_epoch": 16,
|
1179 |
+
"best_eval_accuracy": 0.69,
|
1180 |
+
"epoch": 77.0,
|
1181 |
+
"step": 1925
|
1182 |
+
},
|
1183 |
+
{
|
1184 |
+
"epoch": 78.0,
|
1185 |
+
"eval_accuracy": 0.63,
|
1186 |
+
"eval_loss": 0.2866869568824768,
|
1187 |
+
"eval_runtime": 3.2049,
|
1188 |
+
"eval_samples_per_second": 31.202,
|
1189 |
+
"eval_steps_per_second": 4.056,
|
1190 |
+
"step": 1950
|
1191 |
+
},
|
1192 |
+
{
|
1193 |
+
"best_epoch": 16,
|
1194 |
+
"best_eval_accuracy": 0.69,
|
1195 |
+
"epoch": 78.0,
|
1196 |
+
"step": 1950
|
1197 |
+
},
|
1198 |
+
{
|
1199 |
+
"epoch": 79.0,
|
1200 |
+
"eval_accuracy": 0.62,
|
1201 |
+
"eval_loss": 0.28672027587890625,
|
1202 |
+
"eval_runtime": 3.2024,
|
1203 |
+
"eval_samples_per_second": 31.227,
|
1204 |
+
"eval_steps_per_second": 4.059,
|
1205 |
+
"step": 1975
|
1206 |
+
},
|
1207 |
+
{
|
1208 |
+
"best_epoch": 16,
|
1209 |
+
"best_eval_accuracy": 0.69,
|
1210 |
+
"epoch": 79.0,
|
1211 |
+
"step": 1975
|
1212 |
+
},
|
1213 |
+
{
|
1214 |
+
"epoch": 80.0,
|
1215 |
+
"learning_rate": 0.0,
|
1216 |
+
"loss": 0.4597,
|
1217 |
+
"step": 2000
|
1218 |
+
},
|
1219 |
+
{
|
1220 |
+
"epoch": 80.0,
|
1221 |
+
"eval_accuracy": 0.62,
|
1222 |
+
"eval_loss": 0.2866990268230438,
|
1223 |
+
"eval_runtime": 3.202,
|
1224 |
+
"eval_samples_per_second": 31.23,
|
1225 |
+
"eval_steps_per_second": 4.06,
|
1226 |
+
"step": 2000
|
1227 |
+
},
|
1228 |
+
{
|
1229 |
+
"best_epoch": 16,
|
1230 |
+
"best_eval_accuracy": 0.69,
|
1231 |
+
"epoch": 80.0,
|
1232 |
+
"step": 2000
|
1233 |
+
},
|
1234 |
+
{
|
1235 |
+
"epoch": 80.0,
|
1236 |
+
"step": 2000,
|
1237 |
+
"total_flos": 2.9821702864896e+16,
|
1238 |
+
"train_loss": 0.4624418716430664,
|
1239 |
+
"train_runtime": 1699.8447,
|
1240 |
+
"train_samples_per_second": 18.825,
|
1241 |
+
"train_steps_per_second": 1.177
|
1242 |
+
}
|
1243 |
+
],
|
1244 |
+
"max_steps": 2000,
|
1245 |
+
"num_train_epochs": 80,
|
1246 |
+
"total_flos": 2.9821702864896e+16,
|
1247 |
+
"trial_name": null,
|
1248 |
+
"trial_params": null
|
1249 |
+
}
|