AAA01101312
commited on
Commit
•
98c6233
1
Parent(s):
df5bc1a
Training in progress, step 1500
Browse files
model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 268290900
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8c0b54af839f197fd556e2802b995e04c0e84b6a7dd75ee7bd1cd439b5e136b8
|
3 |
size 268290900
|
run-2/checkpoint-1500/model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 268290900
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8c0b54af839f197fd556e2802b995e04c0e84b6a7dd75ee7bd1cd439b5e136b8
|
3 |
size 268290900
|
run-2/checkpoint-1500/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 536643898
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:735b33efb421584f65e3eeeaef0cc981323128818b267ced3ff6374b95b6fcc7
|
3 |
size 536643898
|
run-2/checkpoint-1500/trainer_state.json
CHANGED
@@ -10,59 +10,59 @@
|
|
10 |
"log_history": [
|
11 |
{
|
12 |
"epoch": 1.0,
|
13 |
-
"eval_accuracy": 0.
|
14 |
-
"eval_loss": 0.
|
15 |
-
"eval_runtime":
|
16 |
-
"eval_samples_per_second":
|
17 |
-
"eval_steps_per_second":
|
18 |
"step": 318
|
19 |
},
|
20 |
{
|
21 |
"epoch": 1.5723270440251573,
|
22 |
-
"grad_norm": 0.
|
23 |
"learning_rate": 1.371069182389937e-05,
|
24 |
-
"loss": 0.
|
25 |
"step": 500
|
26 |
},
|
27 |
{
|
28 |
"epoch": 2.0,
|
29 |
-
"eval_accuracy": 0.
|
30 |
-
"eval_loss": 0.
|
31 |
-
"eval_runtime":
|
32 |
-
"eval_samples_per_second":
|
33 |
-
"eval_steps_per_second":
|
34 |
"step": 636
|
35 |
},
|
36 |
{
|
37 |
"epoch": 3.0,
|
38 |
"eval_accuracy": 0.8664516129032258,
|
39 |
-
"eval_loss": 0.
|
40 |
-
"eval_runtime":
|
41 |
-
"eval_samples_per_second":
|
42 |
-
"eval_steps_per_second":
|
43 |
"step": 954
|
44 |
},
|
45 |
{
|
46 |
"epoch": 3.1446540880503147,
|
47 |
-
"grad_norm": 0.
|
48 |
"learning_rate": 7.421383647798742e-06,
|
49 |
-
"loss": 0.
|
50 |
"step": 1000
|
51 |
},
|
52 |
{
|
53 |
"epoch": 4.0,
|
54 |
-
"eval_accuracy": 0.
|
55 |
-
"eval_loss": 0.
|
56 |
-
"eval_runtime": 15.
|
57 |
-
"eval_samples_per_second":
|
58 |
-
"eval_steps_per_second": 25.
|
59 |
"step": 1272
|
60 |
},
|
61 |
{
|
62 |
"epoch": 4.716981132075472,
|
63 |
-
"grad_norm": 0.
|
64 |
"learning_rate": 1.1320754716981133e-06,
|
65 |
-
"loss": 0.
|
66 |
"step": 1500
|
67 |
}
|
68 |
],
|
@@ -75,8 +75,8 @@
|
|
75 |
"train_batch_size": 48,
|
76 |
"trial_name": null,
|
77 |
"trial_params": {
|
78 |
-
"alpha": 0.
|
79 |
"num_train_epochs": 5,
|
80 |
-
"temperature":
|
81 |
}
|
82 |
}
|
|
|
10 |
"log_history": [
|
11 |
{
|
12 |
"epoch": 1.0,
|
13 |
+
"eval_accuracy": 0.5787096774193549,
|
14 |
+
"eval_loss": 0.21759462356567383,
|
15 |
+
"eval_runtime": 13.1676,
|
16 |
+
"eval_samples_per_second": 235.426,
|
17 |
+
"eval_steps_per_second": 29.466,
|
18 |
"step": 318
|
19 |
},
|
20 |
{
|
21 |
"epoch": 1.5723270440251573,
|
22 |
+
"grad_norm": 0.5513952970504761,
|
23 |
"learning_rate": 1.371069182389937e-05,
|
24 |
+
"loss": 0.3396,
|
25 |
"step": 500
|
26 |
},
|
27 |
{
|
28 |
"epoch": 2.0,
|
29 |
+
"eval_accuracy": 0.8083870967741935,
|
30 |
+
"eval_loss": 0.11128884553909302,
|
31 |
+
"eval_runtime": 13.9571,
|
32 |
+
"eval_samples_per_second": 222.109,
|
33 |
+
"eval_steps_per_second": 27.799,
|
34 |
"step": 636
|
35 |
},
|
36 |
{
|
37 |
"epoch": 3.0,
|
38 |
"eval_accuracy": 0.8664516129032258,
|
39 |
+
"eval_loss": 0.07867251336574554,
|
40 |
+
"eval_runtime": 13.1876,
|
41 |
+
"eval_samples_per_second": 235.07,
|
42 |
+
"eval_steps_per_second": 29.422,
|
43 |
"step": 954
|
44 |
},
|
45 |
{
|
46 |
"epoch": 3.1446540880503147,
|
47 |
+
"grad_norm": 0.5208547711372375,
|
48 |
"learning_rate": 7.421383647798742e-06,
|
49 |
+
"loss": 0.1284,
|
50 |
"step": 1000
|
51 |
},
|
52 |
{
|
53 |
"epoch": 4.0,
|
54 |
+
"eval_accuracy": 0.8825806451612903,
|
55 |
+
"eval_loss": 0.0656495988368988,
|
56 |
+
"eval_runtime": 15.4062,
|
57 |
+
"eval_samples_per_second": 201.217,
|
58 |
+
"eval_steps_per_second": 25.185,
|
59 |
"step": 1272
|
60 |
},
|
61 |
{
|
62 |
"epoch": 4.716981132075472,
|
63 |
+
"grad_norm": 0.4366990029811859,
|
64 |
"learning_rate": 1.1320754716981133e-06,
|
65 |
+
"loss": 0.0913,
|
66 |
"step": 1500
|
67 |
}
|
68 |
],
|
|
|
75 |
"train_batch_size": 48,
|
76 |
"trial_name": null,
|
77 |
"trial_params": {
|
78 |
+
"alpha": 0.8001341619530338,
|
79 |
"num_train_epochs": 5,
|
80 |
+
"temperature": 8
|
81 |
}
|
82 |
}
|
run-2/checkpoint-1500/training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5048
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c3a86d837a7a7b0e3b25e6ad0ff44a7299c5ea5790990e72e2773a24b09c1a55
|
3 |
size 5048
|
runs/May20_08-46-15_MainPC/events.out.tfevents.1716163630.MainPC.215644.3
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9f2a949d1fa504d2e87b204bcda1ea0231e0c2645253873a93b4084c51f8b0f1
|
3 |
+
size 14807
|