gchhablani
commited on
Commit
•
e9503eb
1
Parent(s):
8ee18ba
End of training
Browse files- all_results.json +16 -0
- eval_results.json +11 -0
- runs/Sep19_15-39-26_patrick-general-gpu/events.out.tfevents.1632065982.patrick-general-gpu.509410.0 +2 -2
- runs/Sep19_15-39-26_patrick-general-gpu/events.out.tfevents.1632066425.patrick-general-gpu.509410.2 +3 -0
- train_results.json +8 -0
- trainer_state.json +76 -0
all_results.json
ADDED
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 3.0,
|
3 |
+
"eval_combined_score": 0.8237743551253471,
|
4 |
+
"eval_loss": 0.789396345615387,
|
5 |
+
"eval_pearson": 0.825608960477892,
|
6 |
+
"eval_runtime": 11.1522,
|
7 |
+
"eval_samples": 1500,
|
8 |
+
"eval_samples_per_second": 134.503,
|
9 |
+
"eval_spearmanr": 0.8219397497728022,
|
10 |
+
"eval_steps_per_second": 16.858,
|
11 |
+
"train_loss": 0.9085038361725983,
|
12 |
+
"train_runtime": 428.8873,
|
13 |
+
"train_samples": 5749,
|
14 |
+
"train_samples_per_second": 40.213,
|
15 |
+
"train_steps_per_second": 2.518
|
16 |
+
}
|
eval_results.json
ADDED
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 3.0,
|
3 |
+
"eval_combined_score": 0.8237743551253471,
|
4 |
+
"eval_loss": 0.789396345615387,
|
5 |
+
"eval_pearson": 0.825608960477892,
|
6 |
+
"eval_runtime": 11.1522,
|
7 |
+
"eval_samples": 1500,
|
8 |
+
"eval_samples_per_second": 134.503,
|
9 |
+
"eval_spearmanr": 0.8219397497728022,
|
10 |
+
"eval_steps_per_second": 16.858
|
11 |
+
}
|
runs/Sep19_15-39-26_patrick-general-gpu/events.out.tfevents.1632065982.patrick-general-gpu.509410.0
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 sha256:0638a4ef870de82ad10154791eb11503693287e26b81a3a64e933dd40d661d06
|
3 |
+
size 5250
|
runs/Sep19_15-39-26_patrick-general-gpu/events.out.tfevents.1632066425.patrick-general-gpu.509410.2
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:cf085b82b5e253c3e88f872da1184e011bfe117bc0eebabd71677d6e3ffbdaeb
|
3 |
+
size 473
|
train_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 3.0,
|
3 |
+
"train_loss": 0.9085038361725983,
|
4 |
+
"train_runtime": 428.8873,
|
5 |
+
"train_samples": 5749,
|
6 |
+
"train_samples_per_second": 40.213,
|
7 |
+
"train_steps_per_second": 2.518
|
8 |
+
}
|
trainer_state.json
ADDED
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 3.0,
|
5 |
+
"global_step": 1080,
|
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 |
+
"learning_rate": 1.3333333333333333e-05,
|
13 |
+
"loss": 1.5473,
|
14 |
+
"step": 360
|
15 |
+
},
|
16 |
+
{
|
17 |
+
"epoch": 1.0,
|
18 |
+
"eval_combined_score": 0.8120419393156919,
|
19 |
+
"eval_loss": 0.7750807404518127,
|
20 |
+
"eval_pearson": 0.8115414805311705,
|
21 |
+
"eval_runtime": 10.8984,
|
22 |
+
"eval_samples_per_second": 137.634,
|
23 |
+
"eval_spearmanr": 0.8125423981002132,
|
24 |
+
"eval_steps_per_second": 17.25,
|
25 |
+
"step": 360
|
26 |
+
},
|
27 |
+
{
|
28 |
+
"epoch": 2.0,
|
29 |
+
"learning_rate": 6.666666666666667e-06,
|
30 |
+
"loss": 0.6954,
|
31 |
+
"step": 720
|
32 |
+
},
|
33 |
+
{
|
34 |
+
"epoch": 2.0,
|
35 |
+
"eval_combined_score": 0.8145095741908042,
|
36 |
+
"eval_loss": 0.8717353343963623,
|
37 |
+
"eval_pearson": 0.8159860577723805,
|
38 |
+
"eval_runtime": 10.909,
|
39 |
+
"eval_samples_per_second": 137.502,
|
40 |
+
"eval_spearmanr": 0.8130330906092277,
|
41 |
+
"eval_steps_per_second": 17.234,
|
42 |
+
"step": 720
|
43 |
+
},
|
44 |
+
{
|
45 |
+
"epoch": 3.0,
|
46 |
+
"learning_rate": 0.0,
|
47 |
+
"loss": 0.4828,
|
48 |
+
"step": 1080
|
49 |
+
},
|
50 |
+
{
|
51 |
+
"epoch": 3.0,
|
52 |
+
"eval_combined_score": 0.8237743551253471,
|
53 |
+
"eval_loss": 0.789396345615387,
|
54 |
+
"eval_pearson": 0.825608960477892,
|
55 |
+
"eval_runtime": 10.9264,
|
56 |
+
"eval_samples_per_second": 137.282,
|
57 |
+
"eval_spearmanr": 0.8219397497728022,
|
58 |
+
"eval_steps_per_second": 17.206,
|
59 |
+
"step": 1080
|
60 |
+
},
|
61 |
+
{
|
62 |
+
"epoch": 3.0,
|
63 |
+
"step": 1080,
|
64 |
+
"total_flos": 3067148834731008.0,
|
65 |
+
"train_loss": 0.9085038361725983,
|
66 |
+
"train_runtime": 428.8873,
|
67 |
+
"train_samples_per_second": 40.213,
|
68 |
+
"train_steps_per_second": 2.518
|
69 |
+
}
|
70 |
+
],
|
71 |
+
"max_steps": 1080,
|
72 |
+
"num_train_epochs": 3,
|
73 |
+
"total_flos": 3067148834731008.0,
|
74 |
+
"trial_name": null,
|
75 |
+
"trial_params": null
|
76 |
+
}
|