Commit
·
6b234f3
1
Parent(s):
de727ec
End of training
Browse files
all_results.json
CHANGED
@@ -1,11 +1,11 @@
|
|
1 |
{
|
2 |
"epoch": 5.0,
|
3 |
-
"eval_loss":
|
4 |
-
"eval_runtime": 2.
|
5 |
-
"eval_samples_per_second":
|
6 |
-
"eval_steps_per_second": 1.
|
7 |
-
"train_loss":
|
8 |
-
"train_runtime":
|
9 |
-
"train_samples_per_second": 1.
|
10 |
-
"train_steps_per_second": 0.
|
11 |
}
|
|
|
1 |
{
|
2 |
"epoch": 5.0,
|
3 |
+
"eval_loss": 6787.48388671875,
|
4 |
+
"eval_runtime": 2.1947,
|
5 |
+
"eval_samples_per_second": 14.58,
|
6 |
+
"eval_steps_per_second": 1.823,
|
7 |
+
"train_loss": 6234.263671875,
|
8 |
+
"train_runtime": 98.3256,
|
9 |
+
"train_samples_per_second": 1.627,
|
10 |
+
"train_steps_per_second": 0.203
|
11 |
}
|
eval_results.json
CHANGED
@@ -1,7 +1,7 @@
|
|
1 |
{
|
2 |
"epoch": 5.0,
|
3 |
-
"eval_loss":
|
4 |
-
"eval_runtime": 2.
|
5 |
-
"eval_samples_per_second":
|
6 |
-
"eval_steps_per_second": 1.
|
7 |
}
|
|
|
1 |
{
|
2 |
"epoch": 5.0,
|
3 |
+
"eval_loss": 6787.48388671875,
|
4 |
+
"eval_runtime": 2.1947,
|
5 |
+
"eval_samples_per_second": 14.58,
|
6 |
+
"eval_steps_per_second": 1.823
|
7 |
}
|
runs/Jul19_09-26-30_amy-2-gpu/events.out.tfevents.1689758914.amy-2-gpu.65271.1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e68a853e459590d055a3178d151a8352e199e4bedeb72327642d9dc66ec1591e
|
3 |
+
size 354
|
train_results.json
CHANGED
@@ -1,7 +1,7 @@
|
|
1 |
{
|
2 |
"epoch": 5.0,
|
3 |
-
"train_loss":
|
4 |
-
"train_runtime":
|
5 |
-
"train_samples_per_second": 1.
|
6 |
-
"train_steps_per_second": 0.
|
7 |
}
|
|
|
1 |
{
|
2 |
"epoch": 5.0,
|
3 |
+
"train_loss": 6234.263671875,
|
4 |
+
"train_runtime": 98.3256,
|
5 |
+
"train_samples_per_second": 1.627,
|
6 |
+
"train_steps_per_second": 0.203
|
7 |
}
|
trainer_state.json
CHANGED
@@ -1,5 +1,5 @@
|
|
1 |
{
|
2 |
-
"best_metric":
|
3 |
"best_model_checkpoint": "./coco_outputs/checkpoint-20",
|
4 |
"epoch": 5.0,
|
5 |
"global_step": 20,
|
@@ -9,64 +9,64 @@
|
|
9 |
"log_history": [
|
10 |
{
|
11 |
"epoch": 1.0,
|
12 |
-
"eval_loss":
|
13 |
-
"eval_runtime": 2.
|
14 |
-
"eval_samples_per_second":
|
15 |
-
"eval_steps_per_second": 1.
|
16 |
"step": 4
|
17 |
},
|
18 |
{
|
19 |
"epoch": 2.0,
|
20 |
-
"eval_loss":
|
21 |
-
"eval_runtime": 2.
|
22 |
-
"eval_samples_per_second":
|
23 |
-
"eval_steps_per_second": 1.
|
24 |
"step": 8
|
25 |
},
|
26 |
{
|
27 |
"epoch": 2.5,
|
28 |
"learning_rate": 1e-05,
|
29 |
-
"loss":
|
30 |
"step": 10
|
31 |
},
|
32 |
{
|
33 |
"epoch": 3.0,
|
34 |
-
"eval_loss":
|
35 |
-
"eval_runtime": 2.
|
36 |
-
"eval_samples_per_second": 14.
|
37 |
-
"eval_steps_per_second": 1.
|
38 |
"step": 12
|
39 |
},
|
40 |
{
|
41 |
"epoch": 4.0,
|
42 |
-
"eval_loss":
|
43 |
-
"eval_runtime": 2.
|
44 |
-
"eval_samples_per_second":
|
45 |
-
"eval_steps_per_second": 1.
|
46 |
"step": 16
|
47 |
},
|
48 |
{
|
49 |
"epoch": 5.0,
|
50 |
"learning_rate": 0.0,
|
51 |
-
"loss":
|
52 |
"step": 20
|
53 |
},
|
54 |
{
|
55 |
"epoch": 5.0,
|
56 |
-
"eval_loss":
|
57 |
-
"eval_runtime": 2.
|
58 |
-
"eval_samples_per_second":
|
59 |
-
"eval_steps_per_second": 1.
|
60 |
"step": 20
|
61 |
},
|
62 |
{
|
63 |
"epoch": 5.0,
|
64 |
"step": 20,
|
65 |
"total_flos": 7.65002115072e+16,
|
66 |
-
"train_loss":
|
67 |
-
"train_runtime":
|
68 |
-
"train_samples_per_second": 1.
|
69 |
-
"train_steps_per_second": 0.
|
70 |
}
|
71 |
],
|
72 |
"max_steps": 20,
|
|
|
1 |
{
|
2 |
+
"best_metric": 6787.48388671875,
|
3 |
"best_model_checkpoint": "./coco_outputs/checkpoint-20",
|
4 |
"epoch": 5.0,
|
5 |
"global_step": 20,
|
|
|
9 |
"log_history": [
|
10 |
{
|
11 |
"epoch": 1.0,
|
12 |
+
"eval_loss": 6788.2998046875,
|
13 |
+
"eval_runtime": 2.0726,
|
14 |
+
"eval_samples_per_second": 15.44,
|
15 |
+
"eval_steps_per_second": 1.93,
|
16 |
"step": 4
|
17 |
},
|
18 |
{
|
19 |
"epoch": 2.0,
|
20 |
+
"eval_loss": 6787.97509765625,
|
21 |
+
"eval_runtime": 2.0833,
|
22 |
+
"eval_samples_per_second": 15.361,
|
23 |
+
"eval_steps_per_second": 1.92,
|
24 |
"step": 8
|
25 |
},
|
26 |
{
|
27 |
"epoch": 2.5,
|
28 |
"learning_rate": 1e-05,
|
29 |
+
"loss": 6175.932,
|
30 |
"step": 10
|
31 |
},
|
32 |
{
|
33 |
"epoch": 3.0,
|
34 |
+
"eval_loss": 6787.7900390625,
|
35 |
+
"eval_runtime": 2.1929,
|
36 |
+
"eval_samples_per_second": 14.592,
|
37 |
+
"eval_steps_per_second": 1.824,
|
38 |
"step": 12
|
39 |
},
|
40 |
{
|
41 |
"epoch": 4.0,
|
42 |
+
"eval_loss": 6787.5068359375,
|
43 |
+
"eval_runtime": 2.0906,
|
44 |
+
"eval_samples_per_second": 15.307,
|
45 |
+
"eval_steps_per_second": 1.913,
|
46 |
"step": 16
|
47 |
},
|
48 |
{
|
49 |
"epoch": 5.0,
|
50 |
"learning_rate": 0.0,
|
51 |
+
"loss": 6292.5953,
|
52 |
"step": 20
|
53 |
},
|
54 |
{
|
55 |
"epoch": 5.0,
|
56 |
+
"eval_loss": 6787.48388671875,
|
57 |
+
"eval_runtime": 2.2266,
|
58 |
+
"eval_samples_per_second": 14.372,
|
59 |
+
"eval_steps_per_second": 1.796,
|
60 |
"step": 20
|
61 |
},
|
62 |
{
|
63 |
"epoch": 5.0,
|
64 |
"step": 20,
|
65 |
"total_flos": 7.65002115072e+16,
|
66 |
+
"train_loss": 6234.263671875,
|
67 |
+
"train_runtime": 98.3256,
|
68 |
+
"train_samples_per_second": 1.627,
|
69 |
+
"train_steps_per_second": 0.203
|
70 |
}
|
71 |
],
|
72 |
"max_steps": 20,
|