{ "p2_solo": [ { "qe": 1.4779, "topographic_error": 0.1521, "umatrix_mean": 1.3672, "umatrix_max": 1.6578, "hit_entropy_norm": 0.9582, "neuronios_ativos": 8, "neuronios_total": 8, "grid": [ 4, 2 ], "profundidade_ghsom": 0, "n_amostras": 1400, "epoca": 1, "passos_fase": 58, "lr": 0.002786, "sigma_G": 1.2, "sigma_efetivo": 0.8973, "dW_rel": 27.757012, "loss_total": 33.6331, "qe_batch_max": 2.8499, "gate_gpt2a": { "gate_mass_media": 0.01562, "gate_argmax_fracao_min": 0.0, "gate_argmax_fracao_max": 0.34375 } }, { "qe": 1.4615, "topographic_error": 0.35, "umatrix_mean": 5.7732, "umatrix_max": 20.7626, "hit_entropy_norm": 0.9351, "neuronios_ativos": 10, "neuronios_total": 12, "grid": [ 4, 3 ], "profundidade_ghsom": 0, "n_amostras": 1400, "epoca": 2, "passos_fase": 116, "lr": 0.002545, "sigma_G": 1.2, "sigma_efetivo": 0.6709, "dW_rel": null, "loss_total": 48.4817, "qe_batch_max": 3.2009, "gate_gpt2a": { "gate_mass_media": 0.01562, "gate_argmax_fracao_min": 0.0, "gate_argmax_fracao_max": 0.34375 } }, { "qe": 1.4473, "topographic_error": 0.4986, "umatrix_mean": 2.8361, "umatrix_max": 3.472, "hit_entropy_norm": 0.8154, "neuronios_ativos": 12, "neuronios_total": 12, "grid": [ 4, 3 ], "profundidade_ghsom": 0, "n_amostras": 1400, "epoca": 3, "passos_fase": 174, "lr": 0.002788, "sigma_G": 1.2, "sigma_efetivo": 0.5016, "dW_rel": 0.08848, "loss_total": 37.5368, "qe_batch_max": 8.7618, "gate_gpt2a": { "gate_mass_media": 0.01562, "gate_argmax_fracao_min": 0.0, "gate_argmax_fracao_max": 0.34375 } }, { "qe": 1.644, "topographic_error": 0.4679, "umatrix_mean": 5.2017, "umatrix_max": 6.6437, "hit_entropy_norm": 0.666, "neuronios_ativos": 9, "neuronios_total": 12, "grid": [ 4, 3 ], "profundidade_ghsom": 0, "n_amostras": 1400, "epoca": 4, "passos_fase": 232, "lr": 0.00287, "sigma_G": 1.2, "sigma_efetivo": 0.3751, "dW_rel": 0.076192, "loss_total": 27.9089, "qe_batch_max": 3.1067, "gate_gpt2a": { "gate_mass_media": 0.01562, "gate_argmax_fracao_min": 0.0, "gate_argmax_fracao_max": 0.34375 } }, { "qe": 1.5559, "topographic_error": 0.4707, "umatrix_mean": 6.5535, "umatrix_max": 8.4304, "hit_entropy_norm": 0.7175, "neuronios_ativos": 7, "neuronios_total": 12, "grid": [ 4, 3 ], "profundidade_ghsom": 0, "n_amostras": 1400, "epoca": 5, "passos_fase": 290, "lr": 0.002888, "sigma_G": 1.2, "sigma_efetivo": 0.2805, "dW_rel": 0.037312, "loss_total": 26.0038, "qe_batch_max": 7.9627, "gate_gpt2a": { "gate_mass_media": 0.01562, "gate_argmax_fracao_min": 0.0, "gate_argmax_fracao_max": 0.34375 } } ], "p3_joint2": [ { "qe": 1.8586, "topographic_error": 0.0029, "umatrix_mean": 20.3854, "umatrix_max": 35.5875, "hit_entropy_norm": 0.4363, "neuronios_ativos": 3, "neuronios_total": 12, "grid": [ 4, 3 ], "profundidade_ghsom": 0, "n_amostras": 1400, "epoca": 1, "passos_fase": 58, "lr": 0.002458, "sigma_G": 1.2, "sigma_efetivo": 0.8973, "dW_rel": 0.77101, "loss_total": 51.8539, "bancos": [ "gpt2a", "gpt2g" ] }, { "qe": 1.0976, "topographic_error": 0.3186, "umatrix_mean": 4.8421, "umatrix_max": 12.9399, "hit_entropy_norm": 0.8334, "neuronios_ativos": 9, "neuronios_total": 12, "grid": [ 4, 3 ], "profundidade_ghsom": 0, "n_amostras": 1400, "epoca": 2, "passos_fase": 116, "lr": 0.002719, "sigma_G": 1.2, "sigma_efetivo": 0.6709, "dW_rel": 0.807516, "loss_total": 5.9197, "bancos": [ "gpt2a", "gpt2g" ] }, { "qe": 1.0571, "topographic_error": 0.0871, "umatrix_mean": 3.9495, "umatrix_max": 15.8021, "hit_entropy_norm": 0.9023, "neuronios_ativos": 9, "neuronios_total": 12, "grid": [ 4, 3 ], "profundidade_ghsom": 0, "n_amostras": 1400, "epoca": 3, "passos_fase": 174, "lr": 0.002627, "sigma_G": 1.2, "sigma_efetivo": 0.5016, "dW_rel": 0.263375, "loss_total": 5.3003, "bancos": [ "gpt2a", "gpt2g" ] }, { "qe": 1.3206, "topographic_error": 0.4186, "umatrix_mean": 10.7546, "umatrix_max": 25.5492, "hit_entropy_norm": 0.6588, "neuronios_ativos": 6, "neuronios_total": 15, "grid": [ 5, 3 ], "profundidade_ghsom": 0, "n_amostras": 1400, "epoca": 4, "passos_fase": 232, "lr": 0.002648, "sigma_G": 1.2, "sigma_efetivo": 0.3751, "dW_rel": null, "loss_total": 6.6735, "bancos": [ "gpt2a", "gpt2g" ] }, { "qe": 1.1492, "topographic_error": 0.3143, "umatrix_mean": 10.6559, "umatrix_max": 27.521, "hit_entropy_norm": 0.7831, "neuronios_ativos": 8, "neuronios_total": 15, "grid": [ 5, 3 ], "profundidade_ghsom": 0, "n_amostras": 1400, "epoca": 5, "passos_fase": 290, "lr": 0.002639, "sigma_G": 1.2, "sigma_efetivo": 0.2805, "dW_rel": 0.086108, "loss_total": 6.4104, "bancos": [ "gpt2a", "gpt2g" ] }, { "qe": 1.0888, "topographic_error": 0.3771, "umatrix_mean": 10.6833, "umatrix_max": 27.6425, "hit_entropy_norm": 0.8163, "neuronios_ativos": 8, "neuronios_total": 15, "grid": [ 5, 3 ], "profundidade_ghsom": 0, "n_amostras": 1400, "epoca": 6, "passos_fase": 348, "lr": 0.002618, "sigma_G": 1.2, "sigma_efetivo": 0.25, "dW_rel": 0.031631, "loss_total": 5.9124, "bancos": [ "gpt2a", "gpt2g" ] }, { "qe": 1.867, "topographic_error": 0.585, "umatrix_mean": 11.418, "umatrix_max": 27.7257, "hit_entropy_norm": 0.3601, "neuronios_ativos": 6, "neuronios_total": 20, "grid": [ 5, 4 ], "profundidade_ghsom": 0, "n_amostras": 1400, "epoca": 7, "passos_fase": 406, "lr": 0.002669, "sigma_G": 1.2, "sigma_efetivo": 0.25, "dW_rel": null, "loss_total": 10.4783, "bancos": [ "gpt2a", "gpt2g" ] }, { "qe": 1.4516, "topographic_error": 0.0136, "umatrix_mean": 12.2908, "umatrix_max": 27.7261, "hit_entropy_norm": 0.4463, "neuronios_ativos": 5, "neuronios_total": 20, "grid": [ 5, 4 ], "profundidade_ghsom": 0, "n_amostras": 1400, "epoca": 8, "passos_fase": 464, "lr": 0.002718, "sigma_G": 1.2, "sigma_efetivo": 0.25, "dW_rel": 0.063062, "loss_total": 8.5865, "bancos": [ "gpt2a", "gpt2g" ] }, { "qe": 1.4493, "topographic_error": 0.0136, "umatrix_mean": 12.5316, "umatrix_max": 27.7261, "hit_entropy_norm": 0.5316, "neuronios_ativos": 4, "neuronios_total": 20, "grid": [ 5, 4 ], "profundidade_ghsom": 0, "n_amostras": 1400, "epoca": 9, "passos_fase": 522, "lr": 0.002726, "sigma_G": 1.2, "sigma_efetivo": 0.25, "dW_rel": 0.017085, "loss_total": 8.9184, "bancos": [ "gpt2a", "gpt2g" ] }, { "qe": 1.3948, "topographic_error": 0.015, "umatrix_mean": 12.7577, "umatrix_max": 27.7261, "hit_entropy_norm": 0.5302, "neuronios_ativos": 4, "neuronios_total": 20, "grid": [ 5, 4 ], "profundidade_ghsom": 0, "n_amostras": 1400, "epoca": 10, "passos_fase": 580, "lr": 0.002716, "sigma_G": 1.2, "sigma_efetivo": 0.25, "dW_rel": 0.02078, "loss_total": 8.4981, "bancos": [ "gpt2a", "gpt2g" ] } ], "p4_joint3": [ { "qe": 8.9441, "topographic_error": 0.0, "umatrix_mean": 3.1639, "umatrix_max": 6.0212, "hit_entropy_norm": 0.0326, "neuronios_ativos": 4, "neuronios_total": 20, "grid": [ 5, 4 ], "profundidade_ghsom": 0, "n_amostras": 1400, "epoca": 1, "passos_fase": 58, "lr": 0.002892, "sigma_G": 1.2, "sigma_efetivo": 0.8973, "dW_rel": 0.959528, "loss_total": 24.2915, "bancos": [ "gpt2a", "gpt2g", "dinov2" ] }, { "qe": 5.3624, "topographic_error": 0.7771, "umatrix_mean": 9.6567, "umatrix_max": 16.6179, "hit_entropy_norm": 0.4193, "neuronios_ativos": 4, "neuronios_total": 20, "grid": [ 5, 4 ], "profundidade_ghsom": 0, "n_amostras": 1400, "epoca": 2, "passos_fase": 116, "lr": 0.00251, "sigma_G": 1.2, "sigma_efetivo": 0.6709, "dW_rel": 0.528964, "loss_total": 61.7857, "bancos": [ "gpt2a", "gpt2g", "dinov2" ] }, { "qe": 2.5749, "topographic_error": 0.3279, "umatrix_mean": 9.2223, "umatrix_max": 12.7764, "hit_entropy_norm": 0.6654, "neuronios_ativos": 5, "neuronios_total": 20, "grid": [ 5, 4 ], "profundidade_ghsom": 0, "n_amostras": 1400, "epoca": 3, "passos_fase": 174, "lr": 0.002758, "sigma_G": 1.2, "sigma_efetivo": 0.5016, "dW_rel": 0.202046, "loss_total": 29.7111, "bancos": [ "gpt2a", "gpt2g", "dinov2" ] }, { "qe": 1.987, "topographic_error": 0.46, "umatrix_mean": 11.989, "umatrix_max": 29.2806, "hit_entropy_norm": 0.7674, "neuronios_ativos": 8, "neuronios_total": 24, "grid": [ 6, 4 ], "profundidade_ghsom": 0, "n_amostras": 1400, "epoca": 4, "passos_fase": 232, "lr": 0.00285, "sigma_G": 1.2, "sigma_efetivo": 0.3751, "dW_rel": null, "loss_total": 13.0733, "bancos": [ "gpt2a", "gpt2g", "dinov2" ] }, { "qe": 1.164, "topographic_error": 0.4093, "umatrix_mean": 11.8627, "umatrix_max": 29.2805, "hit_entropy_norm": 0.8072, "neuronios_ativos": 10, "neuronios_total": 24, "grid": [ 6, 4 ], "profundidade_ghsom": 0, "n_amostras": 1400, "epoca": 5, "passos_fase": 290, "lr": 0.002759, "sigma_G": 1.2, "sigma_efetivo": 0.2805, "dW_rel": 0.034128, "loss_total": 10.8068, "bancos": [ "gpt2a", "gpt2g", "dinov2" ] }, { "qe": 1.1454, "topographic_error": 0.4193, "umatrix_mean": 11.8624, "umatrix_max": 29.2803, "hit_entropy_norm": 0.7993, "neuronios_ativos": 10, "neuronios_total": 24, "grid": [ 6, 4 ], "profundidade_ghsom": 0, "n_amostras": 1400, "epoca": 6, "passos_fase": 348, "lr": 0.002768, "sigma_G": 1.2, "sigma_efetivo": 0.25, "dW_rel": 0.012527, "loss_total": 10.5347, "bancos": [ "gpt2a", "gpt2g", "dinov2" ] } ], "verificacao_inicial": { "m1": { "teorema": "T-ATTVAR (Minkowski, combinacao convexa, tau livre)", "bancos": { "gpt2a": { "var_attn": 21.106752, "var_max_especialista": 74.158035, "razao": 0.2846, "ok": true }, "gpt2g": { "var_attn": 2.520774, "var_max_especialista": 3.943501, "razao": 0.6392, "ok": true }, "dinov2": { "var_attn": 0.145641, "var_max_especialista": 11.383606, "razao": 0.0128, "ok": true } }, "passou": true }, "m2": { "teorema": "T-GNN Dirichlet monotono com eta <= 2/lambda_max", "energias": [ 29.024212, 10.711338 ], "monotona": true, "lambda_max_estimado": 0.3514, "eta_seguro_publicado": 0.35, "passou": true }, "m3": { "teorema": "cota de Rademacher sqrt(1+(K-1)rho)/sqrt(K)", "bancos": { "gpt2a": { "rho_bar": 0.9223, "K": 32, "cota_rademacher": 0.9616 }, "gpt2g": { "rho_bar": 0.9583, "K": 32, "cota_rademacher": 0.9796 }, "dinov2": { "rho_bar": 0.9267, "K": 48, "cota_rademacher": 0.9634 } }, "passou": true } }, "verificacao_final": { "m1": { "teorema": "T-ATTVAR (Minkowski, combinacao convexa, tau livre)", "bancos": { "gpt2a": { "var_attn": 21.106752, "var_max_especialista": 74.158035, "razao": 0.2846, "ok": true }, "gpt2g": { "var_attn": 2.520774, "var_max_especialista": 3.943501, "razao": 0.6392, "ok": true }, "dinov2": { "var_attn": 0.145641, "var_max_especialista": 11.383606, "razao": 0.0128, "ok": true } }, "passou": true }, "m2": { "teorema": "T-GNN Dirichlet monotono com eta <= 2/lambda_max", "energias": [ 29.024212, 10.711338 ], "monotona": true, "lambda_max_estimado": 0.4577, "eta_seguro_publicado": 0.35, "passou": true }, "m3": { "teorema": "cota de Rademacher sqrt(1+(K-1)rho)/sqrt(K)", "bancos": { "gpt2a": { "rho_bar": 0.9223, "K": 32, "cota_rademacher": 0.9616 }, "gpt2g": { "rho_bar": 0.9583, "K": 32, "cota_rademacher": 0.9796 }, "dinov2": { "rho_bar": 0.9267, "K": 48, "cota_rademacher": 0.9634 } }, "passou": true } }, "checkpoint_final": "/home/z/my-project/hako_v2_work/ckpt/hako_v2_final.npz" }