{ "s_deltaL_tgw1.0_wr0.0": { "test_acc": 0.29, "mean_bp_cosine": 0.5333309099078178, "mean_rho": 0.5133568346500397, "mean_nudge": -0.06044131517410278, "bp_cosine_per_layer": [ 0.2888738811016083, 0.5792224407196045, 0.602611780166626, 0.6626155376434326 ], "rho_per_layer": [ 0.2811542749404907, 0.5901442170143127, 0.6039299368858337, 0.5781989097595215 ], "final_value_loss": 0.9101862869262696, "s_type": "deltaL", "term_grad_weight": 1.0, "warmup_ratio": 0.0 }, "s_deltaL_tgw1.0_wr0.05": { "test_acc": 0.389, "mean_bp_cosine": 0.7679609805345535, "mean_rho": 0.6908697336912155, "mean_nudge": -0.0764562301337719, "bp_cosine_per_layer": [ 0.6061833500862122, 0.8235390186309814, 0.8219610452651978, 0.8201605081558228 ], "rho_per_layer": [ 0.6085672974586487, 0.739956259727478, 0.7246303558349609, 0.6903250217437744 ], "final_value_loss": 0.7349927394866943, "s_type": "deltaL", "term_grad_weight": 1.0, "warmup_ratio": 0.05 }, "s_deltaL_tgw1.0_wr0.2": { "test_acc": 0.442, "mean_bp_cosine": 0.5583586767315865, "mean_rho": 0.49782785028219223, "mean_nudge": -0.051266513764858246, "bp_cosine_per_layer": [ 0.2746746242046356, 0.6255925893783569, 0.6661019921302795, 0.6670655012130737 ], "rho_per_layer": [ 0.27920976281166077, 0.5733566284179688, 0.594862163066864, 0.5438828468322754 ], "final_value_loss": 0.47215704221725463, "s_type": "deltaL", "term_grad_weight": 1.0, "warmup_ratio": 0.2 }, "s_deltaL_tgw1.0_wr0.5": { "test_acc": 0.6635, "mean_bp_cosine": 0.34044671431183815, "mean_rho": 0.39984575659036636, "mean_nudge": -0.037964228074997663, "bp_cosine_per_layer": [ 0.14777149260044098, 0.3089486062526703, 0.42195770144462585, 0.4831090569496155 ], "rho_per_layer": [ 0.15686997771263123, 0.35711926221847534, 0.5539429187774658, 0.5314508676528931 ], "final_value_loss": 0.3006336982488632, "s_type": "deltaL", "term_grad_weight": 1.0, "warmup_ratio": 0.5 } }