{ "s_eT_tgw0.0_wr0.2": { "test_acc": 0.532, "mean_bp_cosine": 0.11989971622824669, "mean_rho": 0.16079141944646835, "mean_nudge": -0.010690351715311408, "bp_cosine_per_layer": [ 0.03718709945678711, 0.1347142904996872, 0.15199805796146393, 0.15569941699504852 ], "rho_per_layer": [ 0.08883650600910187, 0.1502864807844162, 0.20964613556861877, 0.19439655542373657 ], "final_value_loss": 0.07598549057245255, "s_type": "eT", "term_grad_weight": 0.0, "warmup_ratio": 0.2 }, "s_eT_tgw0.25_wr0.2": { "test_acc": 0.558, "mean_bp_cosine": 0.22698336280882359, "mean_rho": 0.2679573856294155, "mean_nudge": -0.0202574310824275, "bp_cosine_per_layer": [ 0.09379873424768448, 0.2528688311576843, 0.2722504138946533, 0.2890154719352722 ], "rho_per_layer": [ 0.13511960208415985, 0.289243221282959, 0.33815300464630127, 0.30931371450424194 ], "final_value_loss": 0.16253001551628113, "s_type": "eT", "term_grad_weight": 0.25, "warmup_ratio": 0.2 }, "s_eT_tgw1.0_wr0.2": { "test_acc": 0.558, "mean_bp_cosine": 0.45794273912906647, "mean_rho": 0.5322257168591022, "mean_nudge": -0.038075629621744156, "bp_cosine_per_layer": [ 0.17809242010116577, 0.5349531173706055, 0.5522423982620239, 0.5664830207824707 ], "rho_per_layer": [ 0.22843755781650543, 0.5910820364952087, 0.6707864999771118, 0.638596773147583 ], "final_value_loss": 0.38718592133522034, "s_type": "eT", "term_grad_weight": 1.0, "warmup_ratio": 0.2 }, "s_eT_tgw4.0_wr0.2": { "test_acc": 0.3935, "mean_bp_cosine": 0.5738203078508377, "mean_rho": 0.5946865696460009, "mean_nudge": -0.04504407802596688, "bp_cosine_per_layer": [ 0.07111608982086182, 0.6408071517944336, 0.787927508354187, 0.7954304814338684 ], "rho_per_layer": [ 0.10197217017412186, 0.6885993480682373, 0.7956527471542358, 0.7925220131874084 ], "final_value_loss": 1.2652584648132323, "s_type": "eT", "term_grad_weight": 4.0, "warmup_ratio": 0.2 } }