mohameddemes
commited on
Commit
•
c5178a9
1
Parent(s):
e34847b
End of training
Browse files
all_results.json
ADDED
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 3.0,
|
3 |
+
"eval_accuracy": 0.9011764705882352,
|
4 |
+
"eval_loss": 0.2881990969181061,
|
5 |
+
"eval_runtime": 9.7334,
|
6 |
+
"eval_samples_per_second": 43.664,
|
7 |
+
"eval_steps_per_second": 1.438,
|
8 |
+
"total_flos": 2.8493718701027328e+17,
|
9 |
+
"train_loss": 0.9283154063754612,
|
10 |
+
"train_runtime": 296.0589,
|
11 |
+
"train_samples_per_second": 38.709,
|
12 |
+
"train_steps_per_second": 0.304
|
13 |
+
}
|
eval_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 3.0,
|
3 |
+
"eval_accuracy": 0.9011764705882352,
|
4 |
+
"eval_loss": 0.2881990969181061,
|
5 |
+
"eval_runtime": 9.7334,
|
6 |
+
"eval_samples_per_second": 43.664,
|
7 |
+
"eval_steps_per_second": 1.438
|
8 |
+
}
|
runs/Apr01_16-59-30_1be0831098fd/events.out.tfevents.1711991684.1be0831098fd.4267.1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ce5309ffa3cdedc7b5d141ce5c64beeec7afca824298353fc53e8b8e73639cca
|
3 |
+
size 405
|
train_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 3.0,
|
3 |
+
"total_flos": 2.8493718701027328e+17,
|
4 |
+
"train_loss": 0.9283154063754612,
|
5 |
+
"train_runtime": 296.0589,
|
6 |
+
"train_samples_per_second": 38.709,
|
7 |
+
"train_steps_per_second": 0.304
|
8 |
+
}
|
trainer_state.json
ADDED
@@ -0,0 +1,120 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": 0.9011764705882352,
|
3 |
+
"best_model_checkpoint": "swin-tiny-patch4-window7-224-finetuned-eurosat/checkpoint-90",
|
4 |
+
"epoch": 3.0,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 90,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.33,
|
13 |
+
"grad_norm": 7.228083610534668,
|
14 |
+
"learning_rate": 4.938271604938271e-05,
|
15 |
+
"loss": 2.4954,
|
16 |
+
"step": 10
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 0.67,
|
20 |
+
"grad_norm": 13.833573341369629,
|
21 |
+
"learning_rate": 4.3209876543209875e-05,
|
22 |
+
"loss": 1.6703,
|
23 |
+
"step": 20
|
24 |
+
},
|
25 |
+
{
|
26 |
+
"epoch": 1.0,
|
27 |
+
"grad_norm": 12.008746147155762,
|
28 |
+
"learning_rate": 3.7037037037037037e-05,
|
29 |
+
"loss": 0.9086,
|
30 |
+
"step": 30
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 1.0,
|
34 |
+
"eval_accuracy": 0.8117647058823529,
|
35 |
+
"eval_loss": 0.5746737718582153,
|
36 |
+
"eval_runtime": 7.2281,
|
37 |
+
"eval_samples_per_second": 58.799,
|
38 |
+
"eval_steps_per_second": 1.937,
|
39 |
+
"step": 30
|
40 |
+
},
|
41 |
+
{
|
42 |
+
"epoch": 1.33,
|
43 |
+
"grad_norm": 14.395578384399414,
|
44 |
+
"learning_rate": 3.08641975308642e-05,
|
45 |
+
"loss": 0.6799,
|
46 |
+
"step": 40
|
47 |
+
},
|
48 |
+
{
|
49 |
+
"epoch": 1.67,
|
50 |
+
"grad_norm": 10.231263160705566,
|
51 |
+
"learning_rate": 2.4691358024691357e-05,
|
52 |
+
"loss": 0.588,
|
53 |
+
"step": 50
|
54 |
+
},
|
55 |
+
{
|
56 |
+
"epoch": 2.0,
|
57 |
+
"grad_norm": 23.882062911987305,
|
58 |
+
"learning_rate": 1.8518518518518518e-05,
|
59 |
+
"loss": 0.5057,
|
60 |
+
"step": 60
|
61 |
+
},
|
62 |
+
{
|
63 |
+
"epoch": 2.0,
|
64 |
+
"eval_accuracy": 0.8705882352941177,
|
65 |
+
"eval_loss": 0.35231178998947144,
|
66 |
+
"eval_runtime": 7.011,
|
67 |
+
"eval_samples_per_second": 60.619,
|
68 |
+
"eval_steps_per_second": 1.997,
|
69 |
+
"step": 60
|
70 |
+
},
|
71 |
+
{
|
72 |
+
"epoch": 2.33,
|
73 |
+
"grad_norm": 10.384401321411133,
|
74 |
+
"learning_rate": 1.2345679012345678e-05,
|
75 |
+
"loss": 0.5149,
|
76 |
+
"step": 70
|
77 |
+
},
|
78 |
+
{
|
79 |
+
"epoch": 2.67,
|
80 |
+
"grad_norm": 7.370152473449707,
|
81 |
+
"learning_rate": 6.172839506172839e-06,
|
82 |
+
"loss": 0.5327,
|
83 |
+
"step": 80
|
84 |
+
},
|
85 |
+
{
|
86 |
+
"epoch": 3.0,
|
87 |
+
"grad_norm": 9.536111831665039,
|
88 |
+
"learning_rate": 0.0,
|
89 |
+
"loss": 0.4594,
|
90 |
+
"step": 90
|
91 |
+
},
|
92 |
+
{
|
93 |
+
"epoch": 3.0,
|
94 |
+
"eval_accuracy": 0.9011764705882352,
|
95 |
+
"eval_loss": 0.2881990969181061,
|
96 |
+
"eval_runtime": 7.0288,
|
97 |
+
"eval_samples_per_second": 60.465,
|
98 |
+
"eval_steps_per_second": 1.992,
|
99 |
+
"step": 90
|
100 |
+
},
|
101 |
+
{
|
102 |
+
"epoch": 3.0,
|
103 |
+
"step": 90,
|
104 |
+
"total_flos": 2.8493718701027328e+17,
|
105 |
+
"train_loss": 0.9283154063754612,
|
106 |
+
"train_runtime": 296.0589,
|
107 |
+
"train_samples_per_second": 38.709,
|
108 |
+
"train_steps_per_second": 0.304
|
109 |
+
}
|
110 |
+
],
|
111 |
+
"logging_steps": 10,
|
112 |
+
"max_steps": 90,
|
113 |
+
"num_input_tokens_seen": 0,
|
114 |
+
"num_train_epochs": 3,
|
115 |
+
"save_steps": 500,
|
116 |
+
"total_flos": 2.8493718701027328e+17,
|
117 |
+
"train_batch_size": 32,
|
118 |
+
"trial_name": null,
|
119 |
+
"trial_params": null
|
120 |
+
}
|