Upload part2/sweep-full/results.json with huggingface_hub
Browse files
part2/sweep-full/results.json
ADDED
|
@@ -0,0 +1,86 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"adamw-x0.5": {
|
| 3 |
+
"optimizer": "adamw",
|
| 4 |
+
"multiplier": 0.5,
|
| 5 |
+
"lr": 0.001,
|
| 6 |
+
"val_loss": 6.271700595207365,
|
| 7 |
+
"tokens": 925696
|
| 8 |
+
},
|
| 9 |
+
"adamw-x1": {
|
| 10 |
+
"optimizer": "adamw",
|
| 11 |
+
"multiplier": 1,
|
| 12 |
+
"lr": 0.002,
|
| 13 |
+
"val_loss": 6.330227212943578,
|
| 14 |
+
"tokens": 925696
|
| 15 |
+
},
|
| 16 |
+
"adamw-x2": {
|
| 17 |
+
"optimizer": "adamw",
|
| 18 |
+
"multiplier": 2,
|
| 19 |
+
"lr": 0.004,
|
| 20 |
+
"val_loss": 6.7005881056955205,
|
| 21 |
+
"tokens": 925696
|
| 22 |
+
},
|
| 23 |
+
"mars-x0.5": {
|
| 24 |
+
"optimizer": "mars",
|
| 25 |
+
"multiplier": 0.5,
|
| 26 |
+
"lr": 0.002,
|
| 27 |
+
"val_loss": 6.215449964576088,
|
| 28 |
+
"tokens": 925696
|
| 29 |
+
},
|
| 30 |
+
"mars-x1": {
|
| 31 |
+
"optimizer": "mars",
|
| 32 |
+
"multiplier": 1,
|
| 33 |
+
"lr": 0.004,
|
| 34 |
+
"val_loss": 6.294938824393532,
|
| 35 |
+
"tokens": 925696
|
| 36 |
+
},
|
| 37 |
+
"mars-x2": {
|
| 38 |
+
"optimizer": "mars",
|
| 39 |
+
"multiplier": 2,
|
| 40 |
+
"lr": 0.008,
|
| 41 |
+
"val_loss": 6.686863201880172,
|
| 42 |
+
"tokens": 925696
|
| 43 |
+
},
|
| 44 |
+
"lion-x0.5": {
|
| 45 |
+
"optimizer": "lion",
|
| 46 |
+
"multiplier": 0.5,
|
| 47 |
+
"lr": 0.00025,
|
| 48 |
+
"val_loss": 6.580888452266045,
|
| 49 |
+
"tokens": 925696
|
| 50 |
+
},
|
| 51 |
+
"lion-x1": {
|
| 52 |
+
"optimizer": "lion",
|
| 53 |
+
"multiplier": 1,
|
| 54 |
+
"lr": 0.0005,
|
| 55 |
+
"val_loss": 6.70224142074585,
|
| 56 |
+
"tokens": 925696
|
| 57 |
+
},
|
| 58 |
+
"lion-x2": {
|
| 59 |
+
"optimizer": "lion",
|
| 60 |
+
"multiplier": 2,
|
| 61 |
+
"lr": 0.001,
|
| 62 |
+
"val_loss": 6.770781609380669,
|
| 63 |
+
"tokens": 925696
|
| 64 |
+
},
|
| 65 |
+
"scion-x0.5": {
|
| 66 |
+
"optimizer": "scion",
|
| 67 |
+
"multiplier": 0.5,
|
| 68 |
+
"lr": 0.002,
|
| 69 |
+
"val_loss": 6.179808463974903,
|
| 70 |
+
"tokens": 925696
|
| 71 |
+
},
|
| 72 |
+
"scion-x1": {
|
| 73 |
+
"optimizer": "scion",
|
| 74 |
+
"multiplier": 1,
|
| 75 |
+
"lr": 0.004,
|
| 76 |
+
"val_loss": 6.161676105303256,
|
| 77 |
+
"tokens": 925696
|
| 78 |
+
},
|
| 79 |
+
"scion-x2": {
|
| 80 |
+
"optimizer": "scion",
|
| 81 |
+
"multiplier": 2,
|
| 82 |
+
"lr": 0.008,
|
| 83 |
+
"val_loss": 6.304565461727941,
|
| 84 |
+
"tokens": 925696
|
| 85 |
+
}
|
| 86 |
+
}
|