Upload sigma_lambda_v3/history.json with huggingface_hub
Browse files- sigma_lambda_v3/history.json +21 -9
sigma_lambda_v3/history.json
CHANGED
|
@@ -2,39 +2,47 @@
|
|
| 2 |
"epoch": [
|
| 3 |
1,
|
| 4 |
2,
|
| 5 |
-
3
|
|
|
|
| 6 |
],
|
| 7 |
"gen_loss": [
|
| 8 |
30.399562595346868,
|
| 9 |
22.764286764869794,
|
| 10 |
-
15.396741087545925
|
|
|
|
| 11 |
],
|
| 12 |
"ret_loss": [
|
| 13 |
796116674.0579515,
|
| 14 |
5450725.260929077,
|
| 15 |
-
350043.1876737281
|
|
|
|
| 16 |
],
|
| 17 |
"ppl": [
|
| 18 |
48361.553911128496,
|
| 19 |
10405.292656771971,
|
| 20 |
-
1254.1789620978543
|
|
|
|
| 21 |
],
|
| 22 |
"tf_acc": [
|
| 23 |
9.734349599431513e-06,
|
| 24 |
0.39589275342567987,
|
| 25 |
-
0.4354401710649703
|
|
|
|
| 26 |
],
|
| 27 |
"fr_acc": [
|
| 28 |
0.0,
|
| 29 |
0.160561711572741,
|
|
|
|
| 30 |
0.23937194341606105
|
| 31 |
],
|
| 32 |
"mrr": [
|
| 33 |
0.007931923493742943,
|
| 34 |
0.008492429740726948,
|
| 35 |
-
0.00886960793286562
|
|
|
|
| 36 |
],
|
| 37 |
"r1": [
|
|
|
|
| 38 |
0.0010416667209938169,
|
| 39 |
0.0010416667209938169,
|
| 40 |
0.0010416667209938169
|
|
@@ -42,21 +50,25 @@
|
|
| 42 |
"step_var": [
|
| 43 |
0.15913955867290497,
|
| 44 |
0.1598370373249054,
|
| 45 |
-
0.16560138761997223
|
|
|
|
| 46 |
],
|
| 47 |
"pos_slope": [
|
| 48 |
0.4995734989643097,
|
| 49 |
0.4980151951313019,
|
| 50 |
-
0.4959738850593567
|
|
|
|
| 51 |
],
|
| 52 |
"grad_norm": [
|
| 53 |
2887832256709451.0,
|
| 54 |
NaN,
|
|
|
|
| 55 |
NaN
|
| 56 |
],
|
| 57 |
"w_ret": [
|
| 58 |
0.05,
|
| 59 |
0.05,
|
| 60 |
-
0.05
|
|
|
|
| 61 |
]
|
| 62 |
}
|
|
|
|
| 2 |
"epoch": [
|
| 3 |
1,
|
| 4 |
2,
|
| 5 |
+
3,
|
| 6 |
+
4
|
| 7 |
],
|
| 8 |
"gen_loss": [
|
| 9 |
30.399562595346868,
|
| 10 |
22.764286764869794,
|
| 11 |
+
15.396741087545925,
|
| 12 |
+
13.480450591629728
|
| 13 |
],
|
| 14 |
"ret_loss": [
|
| 15 |
796116674.0579515,
|
| 16 |
5450725.260929077,
|
| 17 |
+
350043.1876737281,
|
| 18 |
+
108628.19653386119
|
| 19 |
],
|
| 20 |
"ppl": [
|
| 21 |
48361.553911128496,
|
| 22 |
10405.292656771971,
|
| 23 |
+
1254.1789620978543,
|
| 24 |
+
264.47601817816616
|
| 25 |
],
|
| 26 |
"tf_acc": [
|
| 27 |
9.734349599431513e-06,
|
| 28 |
0.39589275342567987,
|
| 29 |
+
0.4354401710649703,
|
| 30 |
+
0.44354888428129674
|
| 31 |
],
|
| 32 |
"fr_acc": [
|
| 33 |
0.0,
|
| 34 |
0.160561711572741,
|
| 35 |
+
0.23937194341606105,
|
| 36 |
0.23937194341606105
|
| 37 |
],
|
| 38 |
"mrr": [
|
| 39 |
0.007931923493742943,
|
| 40 |
0.008492429740726948,
|
| 41 |
+
0.00886960793286562,
|
| 42 |
+
0.009372583590447903
|
| 43 |
],
|
| 44 |
"r1": [
|
| 45 |
+
0.0010416667209938169,
|
| 46 |
0.0010416667209938169,
|
| 47 |
0.0010416667209938169,
|
| 48 |
0.0010416667209938169
|
|
|
|
| 50 |
"step_var": [
|
| 51 |
0.15913955867290497,
|
| 52 |
0.1598370373249054,
|
| 53 |
+
0.16560138761997223,
|
| 54 |
+
0.17514024674892426
|
| 55 |
],
|
| 56 |
"pos_slope": [
|
| 57 |
0.4995734989643097,
|
| 58 |
0.4980151951313019,
|
| 59 |
+
0.4959738850593567,
|
| 60 |
+
0.49392592906951904
|
| 61 |
],
|
| 62 |
"grad_norm": [
|
| 63 |
2887832256709451.0,
|
| 64 |
NaN,
|
| 65 |
+
NaN,
|
| 66 |
NaN
|
| 67 |
],
|
| 68 |
"w_ret": [
|
| 69 |
0.05,
|
| 70 |
0.05,
|
| 71 |
+
0.05,
|
| 72 |
+
0.083
|
| 73 |
]
|
| 74 |
}
|