Upload sigma_v5_ablation/V5_Norm_result.json with huggingface_hub
Browse files
sigma_v5_ablation/V5_Norm_result.json
ADDED
|
@@ -0,0 +1,60 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"variant": "V5_Norm",
|
| 3 |
+
"best_fr_acc": 0.26766998320307145,
|
| 4 |
+
"final_tf_acc": 0.49702940098057347,
|
| 5 |
+
"final_ppl": 61.42505665318243,
|
| 6 |
+
"mean_step_var": 0.14897080659866332,
|
| 7 |
+
"mean_alpha_grad": 0.0,
|
| 8 |
+
"n_epochs": 5,
|
| 9 |
+
"history": {
|
| 10 |
+
"epoch": [
|
| 11 |
+
1,
|
| 12 |
+
2,
|
| 13 |
+
3,
|
| 14 |
+
4,
|
| 15 |
+
5
|
| 16 |
+
],
|
| 17 |
+
"train_loss": [
|
| 18 |
+
15.23357816140928,
|
| 19 |
+
11.30786015402596,
|
| 20 |
+
12.723735770125273,
|
| 21 |
+
15.796974123327558,
|
| 22 |
+
17.051196255773867
|
| 23 |
+
],
|
| 24 |
+
"ppl": [
|
| 25 |
+
2328.6654053399775,
|
| 26 |
+
73.83668606269272,
|
| 27 |
+
39.19834289015506,
|
| 28 |
+
46.67963716860922,
|
| 29 |
+
61.42505665318243
|
| 30 |
+
],
|
| 31 |
+
"tf_acc": [
|
| 32 |
+
0.45193989363600673,
|
| 33 |
+
0.4750784426338554,
|
| 34 |
+
0.49069233939134355,
|
| 35 |
+
0.49618900213182254,
|
| 36 |
+
0.49702940098057347
|
| 37 |
+
],
|
| 38 |
+
"fr_acc": [
|
| 39 |
+
0.09082589183692125,
|
| 40 |
+
0.24302091903194845,
|
| 41 |
+
0.20578737030942912,
|
| 42 |
+
0.26766998320307145,
|
| 43 |
+
0.22474069027377852
|
| 44 |
+
],
|
| 45 |
+
"step_var": [
|
| 46 |
+
0.14152660965919495,
|
| 47 |
+
0.1455157846212387,
|
| 48 |
+
0.14943934977054596,
|
| 49 |
+
0.15299569070339203,
|
| 50 |
+
0.155376598238945
|
| 51 |
+
],
|
| 52 |
+
"alpha_grad": [
|
| 53 |
+
0.0,
|
| 54 |
+
0.0,
|
| 55 |
+
0.0,
|
| 56 |
+
0.0,
|
| 57 |
+
0.0
|
| 58 |
+
]
|
| 59 |
+
}
|
| 60 |
+
}
|