{ "s_eT_tgw1.0_wr0.0": { "test_acc": 0.4585, "mean_bp_cosine": 0.6755405366420746, "mean_rho": 0.6671112030744553, "mean_nudge": -0.05872981250286102, "bp_cosine_per_layer": [ 0.4783632755279541, 0.7401086091995239, 0.7420414686203003, 0.74164879322052 ], "rho_per_layer": [ 0.49320322275161743, 0.7312472462654114, 0.7255045771598816, 0.7184897661209106 ], "final_value_loss": 0.544479172039032, "s_type": "eT", "term_grad_weight": 1.0, "warmup_ratio": 0.0 }, "s_eT_tgw1.0_wr0.05": { "test_acc": 0.4495, "mean_bp_cosine": 0.45614035055041313, "mean_rho": 0.5053879581391811, "mean_nudge": -0.04452316835522652, "bp_cosine_per_layer": [ 0.12354414165019989, 0.5106832385063171, 0.5928497314453125, 0.597484290599823 ], "rho_per_layer": [ 0.16242454946041107, 0.5767663717269897, 0.6462773084640503, 0.6360836029052734 ], "final_value_loss": 0.6794619485855102, "s_type": "eT", "term_grad_weight": 1.0, "warmup_ratio": 0.05 }, "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_tgw1.0_wr0.5": { "test_acc": 0.663, "mean_bp_cosine": 0.2331971675157547, "mean_rho": 0.3403939865529537, "mean_nudge": -0.029925621580332518, "bp_cosine_per_layer": [ 0.11287420988082886, 0.2289769947528839, 0.2839260697364807, 0.3070113956928253 ], "rho_per_layer": [ 0.1355699747800827, 0.3721342980861664, 0.43592938780784607, 0.4179422855377197 ], "final_value_loss": 0.3164243281364441, "s_type": "eT", "term_grad_weight": 1.0, "warmup_ratio": 0.5 } }