{"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": 1, "device": "cuda", "log_every": 500, "max_steps": 0, "probe_bs": 256, "outdir": "results", "tag": "scale_v3_cifar10_sdil_w64_d20_s1", "n_in": 3072, "n_out": 10}, "provenance": {"git_commit": "d4aef1a571c21e6e4e4125c90aa58da7f14e62e3", "git_dirty": false}, "steps": [{"step": 0, "epoch": 0, "train_loss": 2.8419671058654785, "cos_r_negg": [-0.03832801803946495, -0.0038717780262231827, 0.09431726485490799, -0.009484627284109592, -0.03767269104719162, -0.0004117076750844717, -0.050135478377342224, 0.030684102326631546, -0.06318461894989014, 0.02629108354449272, -0.004418324213474989, -0.061745937913656235, -0.012763015925884247, -0.003970361314713955, -0.01827937737107277, 0.04001510515809059, -0.07552376389503479, 0.06575316190719604, -0.030957911163568497, -0.022568587213754654], "cos_innovation_negg": [-0.03832801803946495, -0.0038717780262231827, 0.09431726485490799, -0.009484627284109592, -0.03767269104719162, -0.0004117076750844717, -0.050135478377342224, 0.030684102326631546, -0.06318461894989014, 0.02629108354449272, -0.004418324213474989, -0.061745937913656235, -0.012763015925884247, -0.003970361314713955, -0.01827937737107277, 0.04001510515809059, -0.07552376389503479, 0.06575316190719604, -0.030957911163568497, -0.022568587213754654], "cos_apical_negg": [-0.03832801803946495, -0.0038717780262231827, 0.09431726485490799, -0.009484627284109592, -0.03767269104719162, -0.0004117076750844717, -0.050135478377342224, 0.030684102326631546, -0.06318461894989014, 0.02629108354449272, -0.004418324213474989, -0.061745937913656235, -0.012763015925884247, -0.003970361314713955, -0.01827937737107277, 0.04001510515809059, -0.07552376389503479, 0.06575316190719604, -0.030957911163568497, -0.022568587213754654], "cos_Ac_negg": [-0.03832801803946495, -0.0038717780262231827, 0.09431726485490799, -0.009484627284109592, -0.03767269104719162, -0.0004117076750844717, -0.050135478377342224, 0.030684102326631546, -0.06318461894989014, 0.02629108354449272, -0.004418324213474989, -0.061745937913656235, -0.012763015925884247, -0.003970361314713955, -0.01827937737107277, 0.04001510515809059, -0.07552376389503479, 0.06575316190719604, -0.030957911163568497, -0.022568587213754654], "r_norm": [2.383727788925171, 2.386284351348877, 2.621631383895874, 2.657865047454834, 2.553990364074707, 2.6697373390197754, 2.3678808212280273, 2.5390899181365967, 2.350024700164795, 2.5546822547912598, 2.4880409240722656, 2.4154088497161865, 2.420438051223755, 2.5935864448547363, 2.4879231452941895, 2.435194969177246, 2.4192099571228027, 2.4645187854766846, 2.576882839202881, 2.6583757400512695], "ldr": {"dL_sdil": -0.008134841918945312, "dL_bp": -0.44547176361083984, "ratio": 0.018261184172498612}}, {"epoch_end": 0, "step": 391, "test_acc": 0.3635, "test_loss": 1.797163232421875}, {"step": 500, "epoch": 1, "train_loss": 1.6951638460159302, "cos_r_negg": [0.6561729907989502, 0.11741368472576141, 0.030840160325169563, 0.034698858857154846, -0.008666852489113808, 0.07916469126939774, 0.05245378986001015, -0.008306588046252728, 0.10829383134841919, 0.08109650760889053, 0.11825691908597946, 0.11434219777584076, 0.032747428864240646, 0.03011179529130459, 0.08245426416397095, 0.07161078602075577, 0.04718869924545288, 0.050318993628025055, 0.04196271672844887, 0.014372127130627632], "cos_innovation_negg": [0.6561729907989502, 0.11741368472576141, 0.030840160325169563, 0.034698858857154846, -0.008666852489113808, 0.07916469126939774, 0.05245378986001015, -0.008306588046252728, 0.10829383134841919, 0.08109650760889053, 0.11825691908597946, 0.11434219777584076, 0.032747428864240646, 0.03011179529130459, 0.08245426416397095, 0.07161078602075577, 0.04718869924545288, 0.050318993628025055, 0.04196271672844887, 0.014372127130627632], "cos_apical_negg": [0.6561729907989502, 0.11741368472576141, 0.030840160325169563, 0.034698858857154846, -0.008666852489113808, 0.07916469126939774, 0.05245378986001015, -0.008306588046252728, 0.10829383134841919, 0.08109650760889053, 0.11825691908597946, 0.11434219777584076, 0.032747428864240646, 0.03011179529130459, 0.08245426416397095, 0.07161078602075577, 0.04718869924545288, 0.050318993628025055, 0.04196271672844887, 0.014372127130627632], "cos_Ac_negg": [0.6561729907989502, 0.11741368472576141, 0.030840160325169563, 0.034698858857154846, -0.008666852489113808, 0.07916469126939774, 0.05245378986001015, -0.008306588046252728, 0.10829383134841919, 0.08109650760889053, 0.11825691908597946, 0.11434219777584076, 0.032747428864240646, 0.03011179529130459, 0.08245426416397095, 0.07161078602075577, 0.04718869924545288, 0.050318993628025055, 0.04196271672844887, 0.014372127130627632], "r_norm": [1.6975085735321045, 1.6651679277420044, 1.8013300895690918, 1.8271446228027344, 1.7554285526275635, 1.8282026052474976, 1.6426830291748047, 1.7468976974487305, 1.6389950513839722, 1.7817060947418213, 1.751086950302124, 1.6610617637634277, 1.688539981842041, 1.8001314401626587, 1.7054901123046875, 1.692521572113037, 1.6640539169311523, 1.7108657360076904, 1.7944279909133911, 1.802899718284607]}, {"epoch_end": 1, "step": 782, "test_acc": 0.3912, "test_loss": 1.7607303344726561}, {"step": 1000, "epoch": 2, "train_loss": 1.7818326950073242, "cos_r_negg": [0.7159899473190308, 0.14886590838432312, 0.1490955650806427, 0.13943776488304138, 0.0706753134727478, 0.10063756257295609, 0.1731964647769928, 0.05207788944244385, 0.19616904854774475, 0.16825328767299652, 0.1567486822605133, 0.1571846455335617, 0.13767178356647491, 0.09135496616363525, 0.18695303797721863, 0.13501423597335815, 0.16141994297504425, 0.10043899714946747, 0.07697904855012894, 0.09337493032217026], "cos_innovation_negg": [0.7159899473190308, 0.14886590838432312, 0.1490955650806427, 0.13943776488304138, 0.0706753134727478, 0.10063756257295609, 0.1731964647769928, 0.05207788944244385, 0.19616904854774475, 0.16825328767299652, 0.1567486822605133, 0.1571846455335617, 0.13767178356647491, 0.09135496616363525, 0.18695303797721863, 0.13501423597335815, 0.16141994297504425, 0.10043899714946747, 0.07697904855012894, 0.09337493032217026], "cos_apical_negg": [0.7159899473190308, 0.14886590838432312, 0.1490955650806427, 0.13943776488304138, 0.0706753134727478, 0.10063756257295609, 0.1731964647769928, 0.05207788944244385, 0.19616904854774475, 0.16825328767299652, 0.1567486822605133, 0.1571846455335617, 0.13767178356647491, 0.09135496616363525, 0.18695303797721863, 0.13501423597335815, 0.16141994297504425, 0.10043899714946747, 0.07697904855012894, 0.09337493032217026], "cos_Ac_negg": [0.7159899473190308, 0.14886590838432312, 0.1490955650806427, 0.13943776488304138, 0.0706753134727478, 0.10063756257295609, 0.1731964647769928, 0.05207788944244385, 0.19616904854774475, 0.16825328767299652, 0.1567486822605133, 0.1571846455335617, 0.13767178356647491, 0.09135496616363525, 0.18695303797721863, 0.13501423597335815, 0.16141994297504425, 0.10043899714946747, 0.07697904855012894, 0.09337493032217026], "r_norm": [1.4526207447052002, 1.3498198986053467, 1.4525245428085327, 1.4825406074523926, 1.4117770195007324, 1.4763624668121338, 1.3297128677368164, 1.4090960025787354, 1.329675555229187, 1.443860411643982, 1.4163719415664673, 1.3448214530944824, 1.3623509407043457, 1.457201600074768, 1.3788962364196777, 1.3688533306121826, 1.3480165004730225, 1.3783578872680664, 1.4392915964126587, 1.4520467519760132]}, {"epoch_end": 2, "step": 1173, "test_acc": 0.4019, "test_loss": 1.728396875}, {"step": 1500, "epoch": 3, "train_loss": 1.6390485763549805, "cos_r_negg": [0.7465548515319824, 0.24588562548160553, 0.2447001039981842, 0.21500159800052643, 0.14729806780815125, 0.18949896097183228, 0.2695710062980652, 0.16623686254024506, 0.2824299931526184, 0.2594683766365051, 0.26125460863113403, 0.24258169531822205, 0.2402794063091278, 0.18759992718696594, 0.27654188871383667, 0.24603475630283356, 0.23048344254493713, 0.17891958355903625, 0.21015974879264832, 0.19055385887622833], "cos_innovation_negg": [0.7465548515319824, 0.24588562548160553, 0.2447001039981842, 0.21500159800052643, 0.14729806780815125, 0.18949896097183228, 0.2695710062980652, 0.16623686254024506, 0.2824299931526184, 0.2594683766365051, 0.26125460863113403, 0.24258169531822205, 0.2402794063091278, 0.18759992718696594, 0.27654188871383667, 0.24603475630283356, 0.23048344254493713, 0.17891958355903625, 0.21015974879264832, 0.19055385887622833], "cos_apical_negg": [0.7465548515319824, 0.24588562548160553, 0.2447001039981842, 0.21500159800052643, 0.14729806780815125, 0.18949896097183228, 0.2695710062980652, 0.16623686254024506, 0.2824299931526184, 0.2594683766365051, 0.26125460863113403, 0.24258169531822205, 0.2402794063091278, 0.18759992718696594, 0.27654188871383667, 0.24603475630283356, 0.23048344254493713, 0.17891958355903625, 0.21015974879264832, 0.19055385887622833], "cos_Ac_negg": [0.7465548515319824, 0.24588562548160553, 0.2447001039981842, 0.21500159800052643, 0.14729806780815125, 0.18949896097183228, 0.2695710062980652, 0.16623686254024506, 0.2824299931526184, 0.2594683766365051, 0.26125460863113403, 0.24258169531822205, 0.2402794063091278, 0.18759992718696594, 0.27654188871383667, 0.24603475630283356, 0.23048344254493713, 0.17891958355903625, 0.21015974879264832, 0.19055385887622833], "r_norm": [1.2614234685897827, 1.0995235443115234, 1.182194709777832, 1.2043302059173584, 1.135392189025879, 1.2062187194824219, 1.0847876071929932, 1.136795997619629, 1.0888444185256958, 1.177617073059082, 1.1473309993743896, 1.0996942520141602, 1.1025092601776123, 1.1759319305419922, 1.1341365575790405, 1.1126022338867188, 1.0974838733673096, 1.1178245544433594, 1.1641536951065063, 1.1819517612457275]}, {"epoch_end": 3, "step": 1564, "test_acc": 0.4093, "test_loss": 1.716443115234375}, {"epoch_end": 4, "step": 1955, "test_acc": 0.42, "test_loss": 1.68951611328125}], "final": {"test_acc": 0.42, "test_loss": 1.68951611328125, "wall_s": 30.51420259475708, "cos_r_negg": [0.7864651679992676, 0.32476919889450073, 0.3332240581512451, 0.30350810289382935, 0.25167158246040344, 0.3030981123447418, 0.3791850507259369, 0.26902854442596436, 0.375348836183548, 0.3657348155975342, 0.38253289461135864, 0.36359965801239014, 0.31708061695098877, 0.3004034757614136, 0.3925201892852783, 0.3441108465194702, 0.33789244294166565, 0.3166937828063965, 0.31062114238739014, 0.2980608642101288], "cos_innovation_negg": [0.7864651679992676, 0.32476919889450073, 0.3332240581512451, 0.30350810289382935, 0.25167158246040344, 0.3030981123447418, 0.3791850507259369, 0.26902854442596436, 0.375348836183548, 0.3657348155975342, 0.38253289461135864, 0.36359965801239014, 0.31708061695098877, 0.3004034757614136, 0.3925201892852783, 0.3441108465194702, 0.33789244294166565, 0.3166937828063965, 0.31062114238739014, 0.2980608642101288], "cos_apical_negg": [0.7864651679992676, 0.32476919889450073, 0.3332240581512451, 0.30350810289382935, 0.25167158246040344, 0.3030981123447418, 0.3791850507259369, 0.26902854442596436, 0.375348836183548, 0.3657348155975342, 0.38253289461135864, 0.36359965801239014, 0.31708061695098877, 0.3004034757614136, 0.3925201892852783, 0.3441108465194702, 0.33789244294166565, 0.3166937828063965, 0.31062114238739014, 0.2980608642101288], "cos_Ac_negg": [0.7864651679992676, 0.32476919889450073, 0.3332240581512451, 0.30350810289382935, 0.25167158246040344, 0.3030981123447418, 0.3791850507259369, 0.26902854442596436, 0.375348836183548, 0.3657348155975342, 0.38253289461135864, 0.36359965801239014, 0.31708061695098877, 0.3004034757614136, 0.3925201892852783, 0.3441108465194702, 0.33789244294166565, 0.3166937828063965, 0.31062114238739014, 0.2980608642101288]}}