{
"arch": "x -> Linear(244,H) -> relu -> Linear(H,H) -> relu -> Linear(H,38)",
"batch": 2048,
"data_contested_rows": 347749,
"data_git_sha": "190c60d4ba94aede74fc1211b530aa645628ac05",
"data_rows": 483777,
"data_seed": 1,
"dtype": "f32-le",
"epochs": 80,
"feature_version": 2,
"features_len": 244,
"git_sha": "190c60d4ba94aede74fc1211b530aa645628ac05",
"hidden": 256,
"lr": 0.001,
"param_floats": 138278,
"param_order": [
"l1.weight",
"l1.bias",
"l2.weight",
"l2.bias",
"l3.weight",
"l3.bias"
],
"param_shapes": {
"l1.bias": [
256
],
"l1.weight": [
256,
244
],
"l2.bias": [
256
],
"l2.weight": [
256,
256
],
"l3.bias": [
38
],
"l3.weight": [
38,
256
]
},
"softmax_len": 38,
"teacher": "american()",
"train_rows": 435399,
"trainer": "pons-trainer 0.1.0",
"val_ce": 0.23473694920539856,
"val_rows": 48378,
"val_top1_constructive": 0.9538427591323853,
"val_top1_contested": 0.948486864566803,
"val_top1_overall": 0.9499979615211487,
"wd": 0.01
}