Upload sigma_lambda_proto_v1/history.json with huggingface_hub
Browse files
sigma_lambda_proto_v1/history.json
CHANGED
|
@@ -1,57 +1,71 @@
|
|
| 1 |
{
|
| 2 |
"epoch": [
|
| 3 |
1,
|
| 4 |
-
2
|
|
|
|
| 5 |
],
|
| 6 |
"gen_loss": [
|
| 7 |
14.378914150180238,
|
| 8 |
-
10.41937110927972
|
|
|
|
| 9 |
],
|
| 10 |
"ret_loss": [
|
|
|
|
| 11 |
0.0,
|
| 12 |
0.0
|
| 13 |
],
|
| 14 |
"ppl": [
|
| 15 |
105.51695237627168,
|
| 16 |
-
36.01809758482572
|
|
|
|
| 17 |
],
|
| 18 |
"tf_acc": [
|
| 19 |
0.44998506448388675,
|
| 20 |
-
0.4889349317509485
|
|
|
|
| 21 |
],
|
| 22 |
"fr_acc": [
|
| 23 |
0.15843538324420675,
|
| 24 |
-
0.21595909319438727
|
|
|
|
| 25 |
],
|
| 26 |
"mrr": [
|
| 27 |
0.01984136737883091,
|
| 28 |
-
0.02093728445470333
|
|
|
|
| 29 |
],
|
| 30 |
"r1": [
|
|
|
|
| 31 |
0.0031250000465661287,
|
| 32 |
0.0031250000465661287
|
| 33 |
],
|
| 34 |
"he_proxy": [
|
| 35 |
0.0,
|
| 36 |
-
0.0
|
|
|
|
| 37 |
],
|
| 38 |
"grad_norm": [
|
| 39 |
69556.54960830766,
|
| 40 |
-
524881.9476341035
|
|
|
|
| 41 |
],
|
| 42 |
"step_var": [
|
| 43 |
0.2371368706226349,
|
| 44 |
-
0.276157945394516
|
|
|
|
| 45 |
],
|
| 46 |
"pos_slope": [
|
| 47 |
0.4982355535030365,
|
| 48 |
-
0.4918489158153534
|
|
|
|
| 49 |
],
|
| 50 |
"beta_mean": [
|
| 51 |
0.09964697062969208,
|
| 52 |
-
0.0983697697520256
|
|
|
|
| 53 |
],
|
| 54 |
"w_ret": [
|
|
|
|
| 55 |
0.0,
|
| 56 |
0.0
|
| 57 |
]
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": [
|
| 3 |
1,
|
| 4 |
+
2,
|
| 5 |
+
3
|
| 6 |
],
|
| 7 |
"gen_loss": [
|
| 8 |
14.378914150180238,
|
| 9 |
+
10.41937110927972,
|
| 10 |
+
9.95253911739046
|
| 11 |
],
|
| 12 |
"ret_loss": [
|
| 13 |
+
0.0,
|
| 14 |
0.0,
|
| 15 |
0.0
|
| 16 |
],
|
| 17 |
"ppl": [
|
| 18 |
105.51695237627168,
|
| 19 |
+
36.01809758482572,
|
| 20 |
+
29.905926998953085
|
| 21 |
],
|
| 22 |
"tf_acc": [
|
| 23 |
0.44998506448388675,
|
| 24 |
+
0.4889349317509485,
|
| 25 |
+
0.5010759556689941
|
| 26 |
],
|
| 27 |
"fr_acc": [
|
| 28 |
0.15843538324420675,
|
| 29 |
+
0.21595909319438727,
|
| 30 |
+
0.22207020430549843
|
| 31 |
],
|
| 32 |
"mrr": [
|
| 33 |
0.01984136737883091,
|
| 34 |
+
0.02093728445470333,
|
| 35 |
+
0.021167319267988205
|
| 36 |
],
|
| 37 |
"r1": [
|
| 38 |
+
0.0031250000465661287,
|
| 39 |
0.0031250000465661287,
|
| 40 |
0.0031250000465661287
|
| 41 |
],
|
| 42 |
"he_proxy": [
|
| 43 |
0.0,
|
| 44 |
+
0.0,
|
| 45 |
+
0.0692528888652218
|
| 46 |
],
|
| 47 |
"grad_norm": [
|
| 48 |
69556.54960830766,
|
| 49 |
+
524881.9476341035,
|
| 50 |
+
129648.56937981857
|
| 51 |
],
|
| 52 |
"step_var": [
|
| 53 |
0.2371368706226349,
|
| 54 |
+
0.276157945394516,
|
| 55 |
+
0.3243665397167206
|
| 56 |
],
|
| 57 |
"pos_slope": [
|
| 58 |
0.4982355535030365,
|
| 59 |
+
0.4918489158153534,
|
| 60 |
+
0.48427340388298035
|
| 61 |
],
|
| 62 |
"beta_mean": [
|
| 63 |
0.09964697062969208,
|
| 64 |
+
0.0983697697520256,
|
| 65 |
+
0.09685240685939789
|
| 66 |
],
|
| 67 |
"w_ret": [
|
| 68 |
+
0.0,
|
| 69 |
0.0,
|
| 70 |
0.0
|
| 71 |
]
|