Upload reports/ablate_optimizers_20260901-161212.json with huggingface_hub
Browse files
reports/ablate_optimizers_20260901-161212.json
ADDED
|
@@ -0,0 +1,164 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"embd64 L2 MoE4": {
|
| 3 |
+
"adamw": {
|
| 4 |
+
"final_val": 2.30124405225118,
|
| 5 |
+
"wall_s": 12.318387031555176,
|
| 6 |
+
"hist": [
|
| 7 |
+
{
|
| 8 |
+
"step": 1,
|
| 9 |
+
"train": 3.4174320697784424,
|
| 10 |
+
"val": 3.4070684631665547
|
| 11 |
+
},
|
| 12 |
+
{
|
| 13 |
+
"step": 50,
|
| 14 |
+
"train": 2.7972233295440674,
|
| 15 |
+
"val": 2.761105451981227
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"step": 100,
|
| 19 |
+
"train": 2.4643044471740723,
|
| 20 |
+
"val": 2.4710233728090922
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"step": 150,
|
| 24 |
+
"train": 2.330726146697998,
|
| 25 |
+
"val": 2.30124405225118
|
| 26 |
+
}
|
| 27 |
+
]
|
| 28 |
+
},
|
| 29 |
+
"gum": {
|
| 30 |
+
"final_val": 2.9501822789510093,
|
| 31 |
+
"wall_s": 13.34013032913208,
|
| 32 |
+
"hist": [
|
| 33 |
+
{
|
| 34 |
+
"step": 1,
|
| 35 |
+
"train": 3.410735845565796,
|
| 36 |
+
"val": 3.4028487265110017
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"step": 50,
|
| 40 |
+
"train": 3.2526257038116455,
|
| 41 |
+
"val": 3.242358034849167
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"step": 100,
|
| 45 |
+
"train": 3.062805414199829,
|
| 46 |
+
"val": 3.0734894990921022
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"step": 150,
|
| 50 |
+
"train": 2.962660551071167,
|
| 51 |
+
"val": 2.9501822789510093
|
| 52 |
+
}
|
| 53 |
+
]
|
| 54 |
+
}
|
| 55 |
+
},
|
| 56 |
+
"embd128 L4 MoE4": {
|
| 57 |
+
"adamw": {
|
| 58 |
+
"final_val": 2.838918560743332,
|
| 59 |
+
"wall_s": 23.61824631690979,
|
| 60 |
+
"hist": [
|
| 61 |
+
{
|
| 62 |
+
"step": 1,
|
| 63 |
+
"train": 3.4587783813476562,
|
| 64 |
+
"val": 3.3166430473327635
|
| 65 |
+
},
|
| 66 |
+
{
|
| 67 |
+
"step": 50,
|
| 68 |
+
"train": 2.8018648624420166,
|
| 69 |
+
"val": 2.8400912046432496
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"step": 100,
|
| 73 |
+
"train": 2.8464834690093994,
|
| 74 |
+
"val": 2.8382199803988137
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"step": 150,
|
| 78 |
+
"train": 2.84476637840271,
|
| 79 |
+
"val": 2.838918560743332
|
| 80 |
+
}
|
| 81 |
+
]
|
| 82 |
+
},
|
| 83 |
+
"gum": {
|
| 84 |
+
"final_val": 2.916172901789347,
|
| 85 |
+
"wall_s": 26.45097541809082,
|
| 86 |
+
"hist": [
|
| 87 |
+
{
|
| 88 |
+
"step": 1,
|
| 89 |
+
"train": 3.4262964725494385,
|
| 90 |
+
"val": 3.4049311498800914
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"step": 50,
|
| 94 |
+
"train": 3.2132856845855713,
|
| 95 |
+
"val": 3.20825012922287
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"step": 100,
|
| 99 |
+
"train": 3.0302419662475586,
|
| 100 |
+
"val": 3.035431573788325
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"step": 150,
|
| 104 |
+
"train": 2.915634870529175,
|
| 105 |
+
"val": 2.916172901789347
|
| 106 |
+
}
|
| 107 |
+
]
|
| 108 |
+
}
|
| 109 |
+
},
|
| 110 |
+
"embd192 L6 MoE8": {
|
| 111 |
+
"adamw": {
|
| 112 |
+
"final_val": 2.8389736115932465,
|
| 113 |
+
"wall_s": 43.53582835197449,
|
| 114 |
+
"hist": [
|
| 115 |
+
{
|
| 116 |
+
"step": 1,
|
| 117 |
+
"train": 3.436568021774292,
|
| 118 |
+
"val": 3.2055450399716694
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"step": 50,
|
| 122 |
+
"train": 4.309718608856201,
|
| 123 |
+
"val": 2.8477032919724783
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"step": 100,
|
| 127 |
+
"train": 2.8408186435699463,
|
| 128 |
+
"val": 2.8455129086971285
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"step": 150,
|
| 132 |
+
"train": 2.8663265705108643,
|
| 133 |
+
"val": 2.8389736115932465
|
| 134 |
+
}
|
| 135 |
+
]
|
| 136 |
+
},
|
| 137 |
+
"gum": {
|
| 138 |
+
"final_val": 2.9717845698197682,
|
| 139 |
+
"wall_s": 54.89743709564209,
|
| 140 |
+
"hist": [
|
| 141 |
+
{
|
| 142 |
+
"step": 1,
|
| 143 |
+
"train": 3.4295766353607178,
|
| 144 |
+
"val": 3.407314678033193
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"step": 50,
|
| 148 |
+
"train": 3.2045648097991943,
|
| 149 |
+
"val": 3.160946196317673
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"step": 100,
|
| 153 |
+
"train": 2.967801332473755,
|
| 154 |
+
"val": 2.9622103214263915
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"step": 150,
|
| 158 |
+
"train": 2.975261926651001,
|
| 159 |
+
"val": 2.9717845698197682
|
| 160 |
+
}
|
| 161 |
+
]
|
| 162 |
+
}
|
| 163 |
+
}
|
| 164 |
+
}
|