summaryrefslogtreecommitdiff
path: root/results/scale_v3_cifar10_sdil_w64_d20_s0.json
blob: b13856134347dbe17244f9d211776ac2fc78e5ab (plain)
1
{"args": {"mode": "sdil", "dataset": "cifar10", "depth": 20, "width": 64, "act": "tanh", "residual": 1, "epochs": 5, "batch_size": 128, "train_examples": 0, "task_seed": 0, "task_train_examples": 50000, "task_test_examples": 10000, "task_levels": 8, "task_n_in": 128, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.02, "eta_A": 0.02, "eta_P": 0.002, "momentum": 0.9, "w_scale": 1.0, "a_scale": 1.0, "feedback_scale": 1.0, "pert_sigma": 0.01, "pert_every": 4, "pert_ndirs": 16, "pert_mode": "simultaneous", "use_residual": 1, "raw_scale_control": "none", "learn_A": 1, "learn_P": 1, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "predictor_mode": "diagonal", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 0, "device": "cuda", "log_every": 500, "max_steps": 0, "probe_bs": 256, "outdir": "results", "tag": "scale_v3_cifar10_sdil_w64_d20_s0", "n_in": 3072, "n_out": 10}, "provenance": {"git_commit": "d4aef1a571c21e6e4e4125c90aa58da7f14e62e3", "git_dirty": false}, "steps": [{"step": 0, "epoch": 0, "train_loss": 2.5764718055725098, "cos_r_negg": [-0.00016129831783473492, -0.06577306985855103, -0.01568448357284069, -0.025263119488954544, -0.007076455745846033, -0.030396265909075737, 0.05402042344212532, -0.008685348555445671, -0.003787924535572529, 0.0635077953338623, 0.04159831255674362, -0.017094656825065613, 0.016450731083750725, 0.028717532753944397, 0.03456263244152069, 0.05013108626008034, 0.02396683394908905, -0.0111713707447052, -0.011928788386285305, -0.06732936203479767], "cos_innovation_negg": [-0.00016129831783473492, -0.06577306985855103, -0.01568448357284069, -0.025263119488954544, -0.007076455745846033, -0.030396265909075737, 0.05402042344212532, -0.008685348555445671, -0.003787924535572529, 0.0635077953338623, 0.04159831255674362, -0.017094656825065613, 0.016450731083750725, 0.028717532753944397, 0.03456263244152069, 0.05013108626008034, 0.02396683394908905, -0.0111713707447052, -0.011928788386285305, -0.06732936203479767], "cos_apical_negg": [-0.00016129831783473492, -0.06577306985855103, -0.01568448357284069, -0.025263119488954544, -0.007076455745846033, -0.030396265909075737, 0.05402042344212532, -0.008685348555445671, -0.003787924535572529, 0.0635077953338623, 0.04159831255674362, -0.017094656825065613, 0.016450731083750725, 0.028717532753944397, 0.03456263244152069, 0.05013108626008034, 0.02396683394908905, -0.0111713707447052, -0.011928788386285305, -0.06732936203479767], "cos_Ac_negg": [-0.00016129831783473492, -0.06577306985855103, -0.01568448357284069, -0.025263119488954544, -0.007076455745846033, -0.030396265909075737, 0.05402042344212532, -0.008685348555445671, -0.003787924535572529, 0.0635077953338623, 0.04159831255674362, -0.017094656825065613, 0.016450731083750725, 0.028717532753944397, 0.03456263244152069, 0.05013108626008034, 0.02396683394908905, -0.0111713707447052, -0.011928788386285305, -0.06732936203479767], "r_norm": [2.5359272956848145, 2.363259792327881, 2.4744482040405273, 2.362381935119629, 2.491180419921875, 2.4103646278381348, 2.4312844276428223, 2.457958698272705, 2.500680446624756, 2.4691591262817383, 2.5564472675323486, 2.464780330657959, 2.522806406021118, 2.422983169555664, 2.51682710647583, 2.417123317718506, 2.4799325466156006, 2.410510540008545, 2.518345355987549, 2.583714485168457], "ldr": {"dL_sdil": 0.17090439796447754, "dL_bp": -0.3899955749511719, "ratio": -0.43822137721915194}}, {"epoch_end": 0, "step": 391, "test_acc": 0.3687, "test_loss": 1.8012673461914062}, {"step": 500, "epoch": 1, "train_loss": 1.7086811065673828, "cos_r_negg": [0.6376374363899231, 0.0940077155828476, 0.03495437651872635, 0.07309380173683167, 0.014917327091097832, 0.1222284585237503, 0.10706004500389099, 0.055318817496299744, 0.0830514132976532, 0.08833376318216324, 0.10074761509895325, 0.07532397657632828, 0.07883947342634201, 0.08912834525108337, 0.0839342251420021, 0.04397605359554291, 0.0804271250963211, 0.0932423323392868, 0.03796664997935295, -0.05875082686543465], "cos_innovation_negg": [0.6376374363899231, 0.0940077155828476, 0.03495437651872635, 0.07309380173683167, 0.014917327091097832, 0.1222284585237503, 0.10706004500389099, 0.055318817496299744, 0.0830514132976532, 0.08833376318216324, 0.10074761509895325, 0.07532397657632828, 0.07883947342634201, 0.08912834525108337, 0.0839342251420021, 0.04397605359554291, 0.0804271250963211, 0.0932423323392868, 0.03796664997935295, -0.05875082686543465], "cos_apical_negg": [0.6376374363899231, 0.0940077155828476, 0.03495437651872635, 0.07309380173683167, 0.014917327091097832, 0.1222284585237503, 0.10706004500389099, 0.055318817496299744, 0.0830514132976532, 0.08833376318216324, 0.10074761509895325, 0.07532397657632828, 0.07883947342634201, 0.08912834525108337, 0.0839342251420021, 0.04397605359554291, 0.0804271250963211, 0.0932423323392868, 0.03796664997935295, -0.05875082686543465], "cos_Ac_negg": [0.6376374363899231, 0.0940077155828476, 0.03495437651872635, 0.07309380173683167, 0.014917327091097832, 0.1222284585237503, 0.10706004500389099, 0.055318817496299744, 0.0830514132976532, 0.08833376318216324, 0.10074761509895325, 0.07532397657632828, 0.07883947342634201, 0.08912834525108337, 0.0839342251420021, 0.04397605359554291, 0.0804271250963211, 0.0932423323392868, 0.03796664997935295, -0.05875082686543465], "r_norm": [1.803837776184082, 1.601464033126831, 1.7251532077789307, 1.6350878477096558, 1.7012262344360352, 1.66953706741333, 1.6960203647613525, 1.7070844173431396, 1.7349114418029785, 1.717332363128662, 1.7752326726913452, 1.6861159801483154, 1.7516231536865234, 1.6887290477752686, 1.7301651239395142, 1.656146764755249, 1.709304928779602, 1.65622079372406, 1.7263681888580322, 1.7397291660308838]}, {"epoch_end": 1, "step": 782, "test_acc": 0.3958, "test_loss": 1.7472452026367187}, {"step": 1000, "epoch": 2, "train_loss": 1.724802851676941, "cos_r_negg": [0.7164188623428345, 0.15302452445030212, 0.13245832920074463, 0.16589584946632385, 0.13177190721035004, 0.18693093955516815, 0.15520745515823364, 0.13805121183395386, 0.1568966507911682, 0.1697784662246704, 0.15145957469940186, 0.1711215376853943, 0.19631679356098175, 0.160292848944664, 0.17174603044986725, 0.1303534209728241, 0.1292436122894287, 0.20074130594730377, 0.12403001636266708, 0.025380346924066544], "cos_innovation_negg": [0.7164188623428345, 0.15302452445030212, 0.13245832920074463, 0.16589584946632385, 0.13177190721035004, 0.18693093955516815, 0.15520745515823364, 0.13805121183395386, 0.1568966507911682, 0.1697784662246704, 0.15145957469940186, 0.1711215376853943, 0.19631679356098175, 0.160292848944664, 0.17174603044986725, 0.1303534209728241, 0.1292436122894287, 0.20074130594730377, 0.12403001636266708, 0.025380346924066544], "cos_apical_negg": [0.7164188623428345, 0.15302452445030212, 0.13245832920074463, 0.16589584946632385, 0.13177190721035004, 0.18693093955516815, 0.15520745515823364, 0.13805121183395386, 0.1568966507911682, 0.1697784662246704, 0.15145957469940186, 0.1711215376853943, 0.19631679356098175, 0.160292848944664, 0.17174603044986725, 0.1303534209728241, 0.1292436122894287, 0.20074130594730377, 0.12403001636266708, 0.025380346924066544], "cos_Ac_negg": [0.7164188623428345, 0.15302452445030212, 0.13245832920074463, 0.16589584946632385, 0.13177190721035004, 0.18693093955516815, 0.15520745515823364, 0.13805121183395386, 0.1568966507911682, 0.1697784662246704, 0.15145957469940186, 0.1711215376853943, 0.19631679356098175, 0.160292848944664, 0.17174603044986725, 0.1303534209728241, 0.1292436122894287, 0.20074130594730377, 0.12403001636266708, 0.025380346924066544], "r_norm": [1.5853397846221924, 1.341728687286377, 1.4289796352386475, 1.3634388446807861, 1.4129139184951782, 1.3950185775756836, 1.4073282480239868, 1.4141154289245605, 1.435158371925354, 1.4301507472991943, 1.4711432456970215, 1.4028112888336182, 1.4592729806900024, 1.4015095233917236, 1.4426298141479492, 1.3715099096298218, 1.4224331378936768, 1.3758559226989746, 1.432018518447876, 1.4325391054153442]}, {"epoch_end": 2, "step": 1173, "test_acc": 0.3985, "test_loss": 1.7286903442382813}, {"step": 1500, "epoch": 3, "train_loss": 1.59889554977417, "cos_r_negg": [0.7755821347236633, 0.27906331419944763, 0.2597314119338989, 0.26492631435394287, 0.2164628803730011, 0.28722715377807617, 0.2511995732784271, 0.24087826907634735, 0.2567095160484314, 0.2860667109489441, 0.2725043296813965, 0.2694406509399414, 0.29494285583496094, 0.26279643177986145, 0.2544870376586914, 0.22036334872245789, 0.2836315631866455, 0.2654222548007965, 0.24759063124656677, 0.12340696901082993], "cos_innovation_negg": [0.7755821347236633, 0.27906331419944763, 0.2597314119338989, 0.26492631435394287, 0.2164628803730011, 0.28722715377807617, 0.2511995732784271, 0.24087826907634735, 0.2567095160484314, 0.2860667109489441, 0.2725043296813965, 0.2694406509399414, 0.29494285583496094, 0.26279643177986145, 0.2544870376586914, 0.22036334872245789, 0.2836315631866455, 0.2654222548007965, 0.24759063124656677, 0.12340696901082993], "cos_apical_negg": [0.7755821347236633, 0.27906331419944763, 0.2597314119338989, 0.26492631435394287, 0.2164628803730011, 0.28722715377807617, 0.2511995732784271, 0.24087826907634735, 0.2567095160484314, 0.2860667109489441, 0.2725043296813965, 0.2694406509399414, 0.29494285583496094, 0.26279643177986145, 0.2544870376586914, 0.22036334872245789, 0.2836315631866455, 0.2654222548007965, 0.24759063124656677, 0.12340696901082993], "cos_Ac_negg": [0.7755821347236633, 0.27906331419944763, 0.2597314119338989, 0.26492631435394287, 0.2164628803730011, 0.28722715377807617, 0.2511995732784271, 0.24087826907634735, 0.2567095160484314, 0.2860667109489441, 0.2725043296813965, 0.2694406509399414, 0.29494285583496094, 0.26279643177986145, 0.2544870376586914, 0.22036334872245789, 0.2836315631866455, 0.2654222548007965, 0.24759063124656677, 0.12340696901082993], "r_norm": [1.376938819885254, 1.1093963384628296, 1.1689517498016357, 1.1224530935287476, 1.1579415798187256, 1.149727463722229, 1.153730869293213, 1.1552687883377075, 1.1794215440750122, 1.1795756816864014, 1.2061586380004883, 1.1563897132873535, 1.2001984119415283, 1.151282548904419, 1.1791470050811768, 1.1174728870391846, 1.1649298667907715, 1.1320585012435913, 1.170107126235962, 1.1637229919433594]}, {"epoch_end": 3, "step": 1564, "test_acc": 0.406, "test_loss": 1.7176787963867188}, {"epoch_end": 4, "step": 1955, "test_acc": 0.4109, "test_loss": 1.6992347045898437}], "final": {"test_acc": 0.4109, "test_loss": 1.6992347045898437, "wall_s": 33.37645173072815, "cos_r_negg": [0.8089088201522827, 0.3648538887500763, 0.3584721088409424, 0.3411136865615845, 0.33868902921676636, 0.3662002980709076, 0.35841116309165955, 0.34857532382011414, 0.366268515586853, 0.3697420358657837, 0.3841099739074707, 0.4268256723880768, 0.388345867395401, 0.3773788809776306, 0.3641743063926697, 0.34929484128952026, 0.351291298866272, 0.37015843391418457, 0.3208814263343811, 0.2668011486530304], "cos_innovation_negg": [0.8089088201522827, 0.3648538887500763, 0.3584721088409424, 0.3411136865615845, 0.33868902921676636, 0.3662002980709076, 0.35841116309165955, 0.34857532382011414, 0.366268515586853, 0.3697420358657837, 0.3841099739074707, 0.4268256723880768, 0.388345867395401, 0.3773788809776306, 0.3641743063926697, 0.34929484128952026, 0.351291298866272, 0.37015843391418457, 0.3208814263343811, 0.2668011486530304], "cos_apical_negg": [0.8089088201522827, 0.3648538887500763, 0.3584721088409424, 0.3411136865615845, 0.33868902921676636, 0.3662002980709076, 0.35841116309165955, 0.34857532382011414, 0.366268515586853, 0.3697420358657837, 0.3841099739074707, 0.4268256723880768, 0.388345867395401, 0.3773788809776306, 0.3641743063926697, 0.34929484128952026, 0.351291298866272, 0.37015843391418457, 0.3208814263343811, 0.2668011486530304], "cos_Ac_negg": [0.8089088201522827, 0.3648538887500763, 0.3584721088409424, 0.3411136865615845, 0.33868902921676636, 0.3662002980709076, 0.35841116309165955, 0.34857532382011414, 0.366268515586853, 0.3697420358657837, 0.3841099739074707, 0.4268256723880768, 0.388345867395401, 0.3773788809776306, 0.3641743063926697, 0.34929484128952026, 0.351291298866272, 0.37015843391418457, 0.3208814263343811, 0.2668011486530304]}}