End of training
Browse files- all_results.json +8 -0
- test_results.json +8 -0
- trainer_state.json +182 -0
all_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 3.23,
|
3 |
+
"eval_accuracy": 0.9884726224783862,
|
4 |
+
"eval_loss": 0.0557892881333828,
|
5 |
+
"eval_runtime": 82.0077,
|
6 |
+
"eval_samples_per_second": 4.231,
|
7 |
+
"eval_steps_per_second": 0.537
|
8 |
+
}
|
test_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 3.23,
|
3 |
+
"eval_accuracy": 0.9884726224783862,
|
4 |
+
"eval_loss": 0.0557892881333828,
|
5 |
+
"eval_runtime": 82.0077,
|
6 |
+
"eval_samples_per_second": 4.231,
|
7 |
+
"eval_steps_per_second": 0.537
|
8 |
+
}
|
trainer_state.json
ADDED
@@ -0,0 +1,182 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": 0.9865771812080537,
|
3 |
+
"best_model_checkpoint": "timesformer-base-finetuned-k400-finetuned-ucf101-subset/checkpoint-76",
|
4 |
+
"epoch": 3.22972972972973,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 148,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.07,
|
13 |
+
"grad_norm": 11.713988304138184,
|
14 |
+
"learning_rate": 3.3333333333333335e-05,
|
15 |
+
"loss": 2.2742,
|
16 |
+
"step": 10
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 0.14,
|
20 |
+
"grad_norm": 5.630847454071045,
|
21 |
+
"learning_rate": 4.81203007518797e-05,
|
22 |
+
"loss": 1.1541,
|
23 |
+
"step": 20
|
24 |
+
},
|
25 |
+
{
|
26 |
+
"epoch": 0.2,
|
27 |
+
"grad_norm": 6.317829608917236,
|
28 |
+
"learning_rate": 4.43609022556391e-05,
|
29 |
+
"loss": 0.4566,
|
30 |
+
"step": 30
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 0.26,
|
34 |
+
"eval_accuracy": 0.9328859060402684,
|
35 |
+
"eval_loss": 0.1591632068157196,
|
36 |
+
"eval_runtime": 34.9604,
|
37 |
+
"eval_samples_per_second": 4.262,
|
38 |
+
"eval_steps_per_second": 0.543,
|
39 |
+
"step": 38
|
40 |
+
},
|
41 |
+
{
|
42 |
+
"epoch": 1.01,
|
43 |
+
"grad_norm": 0.9320641160011292,
|
44 |
+
"learning_rate": 4.0601503759398494e-05,
|
45 |
+
"loss": 0.1565,
|
46 |
+
"step": 40
|
47 |
+
},
|
48 |
+
{
|
49 |
+
"epoch": 1.08,
|
50 |
+
"grad_norm": 0.9729604721069336,
|
51 |
+
"learning_rate": 3.6842105263157895e-05,
|
52 |
+
"loss": 0.0466,
|
53 |
+
"step": 50
|
54 |
+
},
|
55 |
+
{
|
56 |
+
"epoch": 1.15,
|
57 |
+
"grad_norm": 4.047702789306641,
|
58 |
+
"learning_rate": 3.3082706766917295e-05,
|
59 |
+
"loss": 0.0813,
|
60 |
+
"step": 60
|
61 |
+
},
|
62 |
+
{
|
63 |
+
"epoch": 1.22,
|
64 |
+
"grad_norm": 0.10719165951013565,
|
65 |
+
"learning_rate": 2.9323308270676693e-05,
|
66 |
+
"loss": 0.0229,
|
67 |
+
"step": 70
|
68 |
+
},
|
69 |
+
{
|
70 |
+
"epoch": 1.26,
|
71 |
+
"eval_accuracy": 0.9865771812080537,
|
72 |
+
"eval_loss": 0.04105457663536072,
|
73 |
+
"eval_runtime": 35.0758,
|
74 |
+
"eval_samples_per_second": 4.248,
|
75 |
+
"eval_steps_per_second": 0.542,
|
76 |
+
"step": 76
|
77 |
+
},
|
78 |
+
{
|
79 |
+
"epoch": 2.03,
|
80 |
+
"grad_norm": 0.15929466485977173,
|
81 |
+
"learning_rate": 2.556390977443609e-05,
|
82 |
+
"loss": 0.0233,
|
83 |
+
"step": 80
|
84 |
+
},
|
85 |
+
{
|
86 |
+
"epoch": 2.09,
|
87 |
+
"grad_norm": 0.6581839919090271,
|
88 |
+
"learning_rate": 2.1804511278195487e-05,
|
89 |
+
"loss": 0.0125,
|
90 |
+
"step": 90
|
91 |
+
},
|
92 |
+
{
|
93 |
+
"epoch": 2.16,
|
94 |
+
"grad_norm": 4.039840221405029,
|
95 |
+
"learning_rate": 1.8045112781954888e-05,
|
96 |
+
"loss": 0.0106,
|
97 |
+
"step": 100
|
98 |
+
},
|
99 |
+
{
|
100 |
+
"epoch": 2.23,
|
101 |
+
"grad_norm": 0.056705888360738754,
|
102 |
+
"learning_rate": 1.4285714285714285e-05,
|
103 |
+
"loss": 0.007,
|
104 |
+
"step": 110
|
105 |
+
},
|
106 |
+
{
|
107 |
+
"epoch": 2.26,
|
108 |
+
"eval_accuracy": 0.9731543624161074,
|
109 |
+
"eval_loss": 0.046408914029598236,
|
110 |
+
"eval_runtime": 34.7494,
|
111 |
+
"eval_samples_per_second": 4.288,
|
112 |
+
"eval_steps_per_second": 0.547,
|
113 |
+
"step": 114
|
114 |
+
},
|
115 |
+
{
|
116 |
+
"epoch": 3.04,
|
117 |
+
"grad_norm": 0.12897975742816925,
|
118 |
+
"learning_rate": 1.0526315789473684e-05,
|
119 |
+
"loss": 0.0084,
|
120 |
+
"step": 120
|
121 |
+
},
|
122 |
+
{
|
123 |
+
"epoch": 3.11,
|
124 |
+
"grad_norm": 0.052158668637275696,
|
125 |
+
"learning_rate": 6.766917293233083e-06,
|
126 |
+
"loss": 0.004,
|
127 |
+
"step": 130
|
128 |
+
},
|
129 |
+
{
|
130 |
+
"epoch": 3.18,
|
131 |
+
"grad_norm": 0.05669572576880455,
|
132 |
+
"learning_rate": 3.007518796992481e-06,
|
133 |
+
"loss": 0.006,
|
134 |
+
"step": 140
|
135 |
+
},
|
136 |
+
{
|
137 |
+
"epoch": 3.23,
|
138 |
+
"eval_accuracy": 0.9731543624161074,
|
139 |
+
"eval_loss": 0.05629577115178108,
|
140 |
+
"eval_runtime": 34.8619,
|
141 |
+
"eval_samples_per_second": 4.274,
|
142 |
+
"eval_steps_per_second": 0.545,
|
143 |
+
"step": 148
|
144 |
+
},
|
145 |
+
{
|
146 |
+
"epoch": 3.23,
|
147 |
+
"step": 148,
|
148 |
+
"total_flos": 1.0268967416722883e+18,
|
149 |
+
"train_loss": 0.2883293984357167,
|
150 |
+
"train_runtime": 1399.2722,
|
151 |
+
"train_samples_per_second": 0.846,
|
152 |
+
"train_steps_per_second": 0.106
|
153 |
+
},
|
154 |
+
{
|
155 |
+
"epoch": 3.23,
|
156 |
+
"eval_accuracy": 0.9884726224783862,
|
157 |
+
"eval_loss": 0.0557892881333828,
|
158 |
+
"eval_runtime": 82.151,
|
159 |
+
"eval_samples_per_second": 4.224,
|
160 |
+
"eval_steps_per_second": 0.536,
|
161 |
+
"step": 148
|
162 |
+
},
|
163 |
+
{
|
164 |
+
"epoch": 3.23,
|
165 |
+
"eval_accuracy": 0.9884726224783862,
|
166 |
+
"eval_loss": 0.0557892881333828,
|
167 |
+
"eval_runtime": 82.0077,
|
168 |
+
"eval_samples_per_second": 4.231,
|
169 |
+
"eval_steps_per_second": 0.537,
|
170 |
+
"step": 148
|
171 |
+
}
|
172 |
+
],
|
173 |
+
"logging_steps": 10,
|
174 |
+
"max_steps": 148,
|
175 |
+
"num_input_tokens_seen": 0,
|
176 |
+
"num_train_epochs": 9223372036854775807,
|
177 |
+
"save_steps": 500,
|
178 |
+
"total_flos": 1.0268967416722883e+18,
|
179 |
+
"train_batch_size": 8,
|
180 |
+
"trial_name": null,
|
181 |
+
"trial_params": null
|
182 |
+
}
|