Upload sigma_lambda_v5/history.json with huggingface_hub
Browse files- sigma_lambda_v5/history.json +20 -8
sigma_lambda_v5/history.json
CHANGED
|
@@ -1,49 +1,61 @@
|
|
| 1 |
{
|
| 2 |
"epoch": [
|
| 3 |
1,
|
| 4 |
-
2
|
|
|
|
| 5 |
],
|
| 6 |
"gen_loss": [
|
| 7 |
16.586591267879488,
|
| 8 |
-
22.012301083468923
|
|
|
|
| 9 |
],
|
| 10 |
"ret_loss": [
|
|
|
|
| 11 |
0.0,
|
| 12 |
0.0
|
| 13 |
],
|
| 14 |
"ppl": [
|
| 15 |
89.82908441173188,
|
| 16 |
-
160.04577946159412
|
|
|
|
| 17 |
],
|
| 18 |
"tf_acc": [
|
| 19 |
0.46026214386231185,
|
| 20 |
-
0.47643615492864355
|
|
|
|
| 21 |
],
|
| 22 |
"fr_acc": [
|
| 23 |
0.12302286153389094,
|
| 24 |
-
0.11233910085013026
|
|
|
|
| 25 |
],
|
| 26 |
"mrr": [
|
| 27 |
0.006059600971639156,
|
| 28 |
-
0.006043825298547745
|
|
|
|
| 29 |
],
|
| 30 |
"r1": [
|
|
|
|
| 31 |
0.0007812500116415322,
|
| 32 |
0.0007812500116415322
|
| 33 |
],
|
| 34 |
"step_var": [
|
| 35 |
0.16505937278270721,
|
| 36 |
-
0.18193045258522034
|
|
|
|
| 37 |
],
|
| 38 |
"pos_slope": [
|
| 39 |
0.4985616207122803,
|
| 40 |
-
0.4935553967952728
|
|
|
|
| 41 |
],
|
| 42 |
"grad_norm": [
|
|
|
|
| 43 |
NaN,
|
| 44 |
NaN
|
| 45 |
],
|
| 46 |
"w_ret": [
|
|
|
|
| 47 |
0.0,
|
| 48 |
0.0
|
| 49 |
]
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": [
|
| 3 |
1,
|
| 4 |
+
2,
|
| 5 |
+
3
|
| 6 |
],
|
| 7 |
"gen_loss": [
|
| 8 |
16.586591267879488,
|
| 9 |
+
22.012301083468923,
|
| 10 |
+
50.61838954157162
|
| 11 |
],
|
| 12 |
"ret_loss": [
|
| 13 |
+
0.0,
|
| 14 |
0.0,
|
| 15 |
0.0
|
| 16 |
],
|
| 17 |
"ppl": [
|
| 18 |
89.82908441173188,
|
| 19 |
+
160.04577946159412,
|
| 20 |
+
15470.654177938472
|
| 21 |
],
|
| 22 |
"tf_acc": [
|
| 23 |
0.46026214386231185,
|
| 24 |
+
0.47643615492864355,
|
| 25 |
+
0.47478209936121774
|
| 26 |
],
|
| 27 |
"fr_acc": [
|
| 28 |
0.12302286153389094,
|
| 29 |
+
0.11233910085013026,
|
| 30 |
+
0.24578219982631747
|
| 31 |
],
|
| 32 |
"mrr": [
|
| 33 |
0.006059600971639156,
|
| 34 |
+
0.006043825298547745,
|
| 35 |
+
0.006044529378414154
|
| 36 |
],
|
| 37 |
"r1": [
|
| 38 |
+
0.0007812500116415322,
|
| 39 |
0.0007812500116415322,
|
| 40 |
0.0007812500116415322
|
| 41 |
],
|
| 42 |
"step_var": [
|
| 43 |
0.16505937278270721,
|
| 44 |
+
0.18193045258522034,
|
| 45 |
+
0.20552654564380646
|
| 46 |
],
|
| 47 |
"pos_slope": [
|
| 48 |
0.4985616207122803,
|
| 49 |
+
0.4935553967952728,
|
| 50 |
+
0.48772382736206055
|
| 51 |
],
|
| 52 |
"grad_norm": [
|
| 53 |
+
NaN,
|
| 54 |
NaN,
|
| 55 |
NaN
|
| 56 |
],
|
| 57 |
"w_ret": [
|
| 58 |
+
0.0,
|
| 59 |
0.0,
|
| 60 |
0.0
|
| 61 |
]
|