Upload reports/mbg_gum-fix2_20260901-160210_report.json with huggingface_hub
Browse files
reports/mbg_gum-fix2_20260901-160210_report.json
ADDED
|
@@ -0,0 +1,61 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"params": 175072,
|
| 3 |
+
"optimizer": "gum",
|
| 4 |
+
"tokenizer": "char",
|
| 5 |
+
"steps": 60,
|
| 6 |
+
"wall_s": 4.843441486358643,
|
| 7 |
+
"history": [
|
| 8 |
+
{
|
| 9 |
+
"step": 1,
|
| 10 |
+
"train_loss": 3.4199516773223877,
|
| 11 |
+
"val_loss": NaN
|
| 12 |
+
},
|
| 13 |
+
{
|
| 14 |
+
"step": 10,
|
| 15 |
+
"train_loss": NaN,
|
| 16 |
+
"val_loss": NaN
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"step": 20,
|
| 20 |
+
"train_loss": NaN,
|
| 21 |
+
"val_loss": NaN
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"step": 30,
|
| 25 |
+
"train_loss": NaN,
|
| 26 |
+
"val_loss": NaN
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"step": 40,
|
| 30 |
+
"train_loss": NaN,
|
| 31 |
+
"val_loss": NaN
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"step": 50,
|
| 35 |
+
"train_loss": NaN,
|
| 36 |
+
"val_loss": NaN
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"step": 60,
|
| 40 |
+
"train_loss": NaN,
|
| 41 |
+
"val_loss": NaN
|
| 42 |
+
}
|
| 43 |
+
],
|
| 44 |
+
"routing": {
|
| 45 |
+
"CAUSAL": {
|
| 46 |
+
"matched": NaN,
|
| 47 |
+
"mismatched": 0.0,
|
| 48 |
+
"delta": NaN
|
| 49 |
+
},
|
| 50 |
+
"SPATIAL": {
|
| 51 |
+
"matched": NaN,
|
| 52 |
+
"mismatched": 0.0,
|
| 53 |
+
"delta": NaN
|
| 54 |
+
},
|
| 55 |
+
"TEMPORAL": {
|
| 56 |
+
"matched": NaN,
|
| 57 |
+
"mismatched": 0.0,
|
| 58 |
+
"delta": NaN
|
| 59 |
+
}
|
| 60 |
+
}
|
| 61 |
+
}
|