summaryrefslogtreecommitdiff
path: root/results/scale_v3_cifar10_sdil_w64_d20_s4.json
blob: 733bd49e45708bf53f7c413a003f2defcf161bb9 (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": 4, "device": "cuda", "log_every": 500, "max_steps": 0, "probe_bs": 256, "outdir": "results", "tag": "scale_v3_cifar10_sdil_w64_d20_s4", "n_in": 3072, "n_out": 10}, "provenance": {"git_commit": "d4aef1a571c21e6e4e4125c90aa58da7f14e62e3", "git_dirty": false}, "steps": [{"step": 0, "epoch": 0, "train_loss": 2.6019721031188965, "cos_r_negg": [-0.005426664371043444, -0.06576360762119293, -0.013913685455918312, -0.0007375492714345455, 0.042505212128162384, 0.028738070279359818, 0.05010166019201279, 0.029932411387562752, -0.03876534104347229, 0.06078220158815384, -0.026165293529629707, -0.01946580968797207, 0.06799495220184326, 0.028784606605768204, -0.024993818253278732, 0.004779389128088951, -0.07169294357299805, 0.02718537673354149, -0.004481066484004259, -0.005686664022505283], "cos_innovation_negg": [-0.005426664371043444, -0.06576360762119293, -0.013913685455918312, -0.0007375492714345455, 0.042505212128162384, 0.028738070279359818, 0.05010166019201279, 0.029932411387562752, -0.03876534104347229, 0.06078220158815384, -0.026165293529629707, -0.01946580968797207, 0.06799495220184326, 0.028784606605768204, -0.024993818253278732, 0.004779389128088951, -0.07169294357299805, 0.02718537673354149, -0.004481066484004259, -0.005686664022505283], "cos_apical_negg": [-0.005426664371043444, -0.06576360762119293, -0.013913685455918312, -0.0007375492714345455, 0.042505212128162384, 0.028738070279359818, 0.05010166019201279, 0.029932411387562752, -0.03876534104347229, 0.06078220158815384, -0.026165293529629707, -0.01946580968797207, 0.06799495220184326, 0.028784606605768204, -0.024993818253278732, 0.004779389128088951, -0.07169294357299805, 0.02718537673354149, -0.004481066484004259, -0.005686664022505283], "cos_Ac_negg": [-0.005426664371043444, -0.06576360762119293, -0.013913685455918312, -0.0007375492714345455, 0.042505212128162384, 0.028738070279359818, 0.05010166019201279, 0.029932411387562752, -0.03876534104347229, 0.06078220158815384, -0.026165293529629707, -0.01946580968797207, 0.06799495220184326, 0.028784606605768204, -0.024993818253278732, 0.004779389128088951, -0.07169294357299805, 0.02718537673354149, -0.004481066484004259, -0.005686664022505283], "r_norm": [2.310703992843628, 2.434225082397461, 2.397228240966797, 2.403865337371826, 2.3869447708129883, 2.3313851356506348, 2.4008541107177734, 2.3483524322509766, 2.427638530731201, 2.4909186363220215, 2.599492073059082, 2.4605350494384766, 2.3516464233398438, 2.5008761882781982, 2.396209716796875, 2.4611973762512207, 2.5358726978302, 2.5047860145568848, 2.4141249656677246, 2.3651657104492188], "ldr": {"dL_sdil": -0.009997129440307617, "dL_bp": -0.33337926864624023, "ratio": 0.029987255898974035}}, {"epoch_end": 0, "step": 391, "test_acc": 0.368, "test_loss": 1.79370205078125}, {"step": 500, "epoch": 1, "train_loss": 1.7215092182159424, "cos_r_negg": [0.6435320973396301, 0.02979099005460739, 0.038289934396743774, 0.11927058547735214, 0.03058742731809616, 0.008062007836997509, 0.024464286863803864, 0.05901014059782028, 0.03652437403798103, 0.05225740000605583, 0.0661177933216095, 0.061135128140449524, 0.0898527055978775, 0.15285257995128632, 0.11218602955341339, 0.0787072628736496, 0.09844708442687988, 0.03050638921558857, 0.048210304230451584, 0.07648247480392456], "cos_innovation_negg": [0.6435320973396301, 0.02979099005460739, 0.038289934396743774, 0.11927058547735214, 0.03058742731809616, 0.008062007836997509, 0.024464286863803864, 0.05901014059782028, 0.03652437403798103, 0.05225740000605583, 0.0661177933216095, 0.061135128140449524, 0.0898527055978775, 0.15285257995128632, 0.11218602955341339, 0.0787072628736496, 0.09844708442687988, 0.03050638921558857, 0.048210304230451584, 0.07648247480392456], "cos_apical_negg": [0.6435320973396301, 0.02979099005460739, 0.038289934396743774, 0.11927058547735214, 0.03058742731809616, 0.008062007836997509, 0.024464286863803864, 0.05901014059782028, 0.03652437403798103, 0.05225740000605583, 0.0661177933216095, 0.061135128140449524, 0.0898527055978775, 0.15285257995128632, 0.11218602955341339, 0.0787072628736496, 0.09844708442687988, 0.03050638921558857, 0.048210304230451584, 0.07648247480392456], "cos_Ac_negg": [0.6435320973396301, 0.02979099005460739, 0.038289934396743774, 0.11927058547735214, 0.03058742731809616, 0.008062007836997509, 0.024464286863803864, 0.05901014059782028, 0.03652437403798103, 0.05225740000605583, 0.0661177933216095, 0.061135128140449524, 0.0898527055978775, 0.15285257995128632, 0.11218602955341339, 0.0787072628736496, 0.09844708442687988, 0.03050638921558857, 0.048210304230451584, 0.07648247480392456], "r_norm": [1.7069836854934692, 1.7300031185150146, 1.7195383310317993, 1.7282086610794067, 1.7062828540802002, 1.6588754653930664, 1.7133336067199707, 1.6682157516479492, 1.7346954345703125, 1.773029088973999, 1.8574137687683105, 1.7484214305877686, 1.6879682540893555, 1.7908145189285278, 1.7090635299682617, 1.7585926055908203, 1.816706895828247, 1.7894766330718994, 1.7323718070983887, 1.6724505424499512]}, {"epoch_end": 1, "step": 782, "test_acc": 0.3892, "test_loss": 1.7542832763671874}, {"step": 1000, "epoch": 2, "train_loss": 1.7330005168914795, "cos_r_negg": [0.732060968875885, 0.13029882311820984, 0.08779944479465485, 0.149687260389328, 0.10239934176206589, 0.12168541550636292, 0.11639503389596939, 0.1499321162700653, 0.07639817893505096, 0.09248349815607071, 0.16656702756881714, 0.19601047039031982, 0.13426366448402405, 0.22377607226371765, 0.15169459581375122, 0.18203602731227875, 0.1945788562297821, 0.10510239750146866, 0.10398594290018082, 0.18568867444992065], "cos_innovation_negg": [0.732060968875885, 0.13029882311820984, 0.08779944479465485, 0.149687260389328, 0.10239934176206589, 0.12168541550636292, 0.11639503389596939, 0.1499321162700653, 0.07639817893505096, 0.09248349815607071, 0.16656702756881714, 0.19601047039031982, 0.13426366448402405, 0.22377607226371765, 0.15169459581375122, 0.18203602731227875, 0.1945788562297821, 0.10510239750146866, 0.10398594290018082, 0.18568867444992065], "cos_apical_negg": [0.732060968875885, 0.13029882311820984, 0.08779944479465485, 0.149687260389328, 0.10239934176206589, 0.12168541550636292, 0.11639503389596939, 0.1499321162700653, 0.07639817893505096, 0.09248349815607071, 0.16656702756881714, 0.19601047039031982, 0.13426366448402405, 0.22377607226371765, 0.15169459581375122, 0.18203602731227875, 0.1945788562297821, 0.10510239750146866, 0.10398594290018082, 0.18568867444992065], "cos_Ac_negg": [0.732060968875885, 0.13029882311820984, 0.08779944479465485, 0.149687260389328, 0.10239934176206589, 0.12168541550636292, 0.11639503389596939, 0.1499321162700653, 0.07639817893505096, 0.09248349815607071, 0.16656702756881714, 0.19601047039031982, 0.13426366448402405, 0.22377607226371765, 0.15169459581375122, 0.18203602731227875, 0.1945788562297821, 0.10510239750146866, 0.10398594290018082, 0.18568867444992065], "r_norm": [1.4766380786895752, 1.4193503856658936, 1.397601842880249, 1.4075932502746582, 1.3944990634918213, 1.3567073345184326, 1.4010192155838013, 1.372812271118164, 1.4164788722991943, 1.4462597370147705, 1.523186206817627, 1.4336186647415161, 1.3892886638641357, 1.4774219989776611, 1.4066295623779297, 1.4459102153778076, 1.4862170219421387, 1.4584931135177612, 1.4115922451019287, 1.375382661819458]}, {"epoch_end": 2, "step": 1173, "test_acc": 0.3939, "test_loss": 1.7306015258789063}, {"step": 1500, "epoch": 3, "train_loss": 1.6213781833648682, "cos_r_negg": [0.7830208539962769, 0.23410356044769287, 0.18149475753307343, 0.24196097254753113, 0.19742465019226074, 0.24159511923789978, 0.24094395339488983, 0.261264443397522, 0.20376703143119812, 0.19569405913352966, 0.2428334653377533, 0.3185262084007263, 0.27173542976379395, 0.3099714517593384, 0.25454744696617126, 0.2608778178691864, 0.3009737730026245, 0.21424521505832672, 0.1892642080783844, 0.27473703026771545], "cos_innovation_negg": [0.7830208539962769, 0.23410356044769287, 0.18149475753307343, 0.24196097254753113, 0.19742465019226074, 0.24159511923789978, 0.24094395339488983, 0.261264443397522, 0.20376703143119812, 0.19569405913352966, 0.2428334653377533, 0.3185262084007263, 0.27173542976379395, 0.3099714517593384, 0.25454744696617126, 0.2608778178691864, 0.3009737730026245, 0.21424521505832672, 0.1892642080783844, 0.27473703026771545], "cos_apical_negg": [0.7830208539962769, 0.23410356044769287, 0.18149475753307343, 0.24196097254753113, 0.19742465019226074, 0.24159511923789978, 0.24094395339488983, 0.261264443397522, 0.20376703143119812, 0.19569405913352966, 0.2428334653377533, 0.3185262084007263, 0.27173542976379395, 0.3099714517593384, 0.25454744696617126, 0.2608778178691864, 0.3009737730026245, 0.21424521505832672, 0.1892642080783844, 0.27473703026771545], "cos_Ac_negg": [0.7830208539962769, 0.23410356044769287, 0.18149475753307343, 0.24196097254753113, 0.19742465019226074, 0.24159511923789978, 0.24094395339488983, 0.261264443397522, 0.20376703143119812, 0.19569405913352966, 0.2428334653377533, 0.3185262084007263, 0.27173542976379395, 0.3099714517593384, 0.25454744696617126, 0.2608778178691864, 0.3009737730026245, 0.21424521505832672, 0.1892642080783844, 0.27473703026771545], "r_norm": [1.2948837280273438, 1.1682419776916504, 1.1449615955352783, 1.1604783535003662, 1.1413787603378296, 1.1188760995864868, 1.1491694450378418, 1.1398184299468994, 1.1714365482330322, 1.1879514455795288, 1.2581785917282104, 1.1974430084228516, 1.1568479537963867, 1.2337846755981445, 1.1641652584075928, 1.2003787755966187, 1.2375003099441528, 1.19940185546875, 1.1574232578277588, 1.147538185119629]}, {"epoch_end": 3, "step": 1564, "test_acc": 0.4134, "test_loss": 1.70615068359375}, {"epoch_end": 4, "step": 1955, "test_acc": 0.4175, "test_loss": 1.685459912109375}], "final": {"test_acc": 0.4175, "test_loss": 1.685459912109375, "wall_s": 30.241274118423462, "cos_r_negg": [0.8183760643005371, 0.32777467370033264, 0.2825019657611847, 0.34088966250419617, 0.28872132301330566, 0.3355516791343689, 0.31173521280288696, 0.3617444336414337, 0.3125666677951813, 0.319580614566803, 0.3575560450553894, 0.4096429944038391, 0.384283185005188, 0.4019993245601654, 0.3340301513671875, 0.3886262774467468, 0.40773576498031616, 0.31151083111763, 0.32826805114746094, 0.39940357208251953], "cos_innovation_negg": [0.8183760643005371, 0.32777467370033264, 0.2825019657611847, 0.34088966250419617, 0.28872132301330566, 0.3355516791343689, 0.31173521280288696, 0.3617444336414337, 0.3125666677951813, 0.319580614566803, 0.3575560450553894, 0.4096429944038391, 0.384283185005188, 0.4019993245601654, 0.3340301513671875, 0.3886262774467468, 0.40773576498031616, 0.31151083111763, 0.32826805114746094, 0.39940357208251953], "cos_apical_negg": [0.8183760643005371, 0.32777467370033264, 0.2825019657611847, 0.34088966250419617, 0.28872132301330566, 0.3355516791343689, 0.31173521280288696, 0.3617444336414337, 0.3125666677951813, 0.319580614566803, 0.3575560450553894, 0.4096429944038391, 0.384283185005188, 0.4019993245601654, 0.3340301513671875, 0.3886262774467468, 0.40773576498031616, 0.31151083111763, 0.32826805114746094, 0.39940357208251953], "cos_Ac_negg": [0.8183760643005371, 0.32777467370033264, 0.2825019657611847, 0.34088966250419617, 0.28872132301330566, 0.3355516791343689, 0.31173521280288696, 0.3617444336414337, 0.3125666677951813, 0.319580614566803, 0.3575560450553894, 0.4096429944038391, 0.384283185005188, 0.4019993245601654, 0.3340301513671875, 0.3886262774467468, 0.40773576498031616, 0.31151083111763, 0.32826805114746094, 0.39940357208251953]}}