diff options
Diffstat (limited to 'results/scale_v3_cifar10_sdil_w64_d10_s2.json')
| -rw-r--r-- | results/scale_v3_cifar10_sdil_w64_d10_s2.json | 1 |
1 files changed, 1 insertions, 0 deletions
diff --git a/results/scale_v3_cifar10_sdil_w64_d10_s2.json b/results/scale_v3_cifar10_sdil_w64_d10_s2.json new file mode 100644 index 0000000..83ec4d5 --- /dev/null +++ b/results/scale_v3_cifar10_sdil_w64_d10_s2.json @@ -0,0 +1 @@ +{"args": {"mode": "sdil", "dataset": "cifar10", "depth": 10, "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": 2, "device": "cuda", "log_every": 500, "max_steps": 0, "probe_bs": 256, "outdir": "results", "tag": "scale_v3_cifar10_sdil_w64_d10_s2", "n_in": 3072, "n_out": 10}, "provenance": {"git_commit": "d4aef1a571c21e6e4e4125c90aa58da7f14e62e3", "git_dirty": false}, "steps": [{"step": 0, "epoch": 0, "train_loss": 2.5496790409088135, "cos_r_negg": [0.014987371861934662, 0.06408058106899261, -0.007605401333421469, 0.025352299213409424, 0.006185218691825867, -0.021754547953605652, -0.034144096076488495, 0.019105970859527588, 0.06043533980846405, -0.025133628398180008], "cos_innovation_negg": [0.014987371861934662, 0.06408058106899261, -0.007605401333421469, 0.025352299213409424, 0.006185218691825867, -0.021754547953605652, -0.034144096076488495, 0.019105970859527588, 0.06043533980846405, -0.025133628398180008], "cos_apical_negg": [0.014987371861934662, 0.06408058106899261, -0.007605401333421469, 0.025352299213409424, 0.006185218691825867, -0.021754547953605652, -0.034144096076488495, 0.019105970859527588, 0.06043533980846405, -0.025133628398180008], "cos_Ac_negg": [0.014987371861934662, 0.06408058106899261, -0.007605401333421469, 0.025352299213409424, 0.006185218691825867, -0.021754547953605652, -0.034144096076488495, 0.019105970859527588, 0.06043533980846405, -0.025133628398180008], "r_norm": [2.45066237449646, 2.2925515174865723, 2.3728554248809814, 2.485140800476074, 2.457198143005371, 2.378668785095215, 2.316483497619629, 2.5695598125457764, 2.4922373294830322, 2.3410820960998535], "ldr": {"dL_sdil": 0.06643986701965332, "dL_bp": -0.34011101722717285, "ratio": -0.19534758844720296}}, {"epoch_end": 0, "step": 391, "test_acc": 0.3608, "test_loss": 1.8144536865234375}, {"step": 500, "epoch": 1, "train_loss": 1.691274881362915, "cos_r_negg": [0.6732789278030396, 0.039163559675216675, 0.04456915706396103, 0.07367390394210815, 0.0845639631152153, 0.07144875824451447, 0.06571881473064423, -0.02073337323963642, 0.08263254165649414, 0.055007077753543854], "cos_innovation_negg": [0.6732789278030396, 0.039163559675216675, 0.04456915706396103, 0.07367390394210815, 0.0845639631152153, 0.07144875824451447, 0.06571881473064423, -0.02073337323963642, 0.08263254165649414, 0.055007077753543854], "cos_apical_negg": [0.6732789278030396, 0.039163559675216675, 0.04456915706396103, 0.07367390394210815, 0.0845639631152153, 0.07144875824451447, 0.06571881473064423, -0.02073337323963642, 0.08263254165649414, 0.055007077753543854], "cos_Ac_negg": [0.6732789278030396, 0.039163559675216675, 0.04456915706396103, 0.07367390394210815, 0.0845639631152153, 0.07144875824451447, 0.06571881473064423, -0.02073337323963642, 0.08263254165649414, 0.055007077753543854], "r_norm": [1.7756133079528809, 1.6064691543579102, 1.6789015531539917, 1.7587956190109253, 1.7500337362289429, 1.6711909770965576, 1.6234984397888184, 1.8019797801971436, 1.7553244829177856, 1.6231260299682617]}, {"epoch_end": 1, "step": 782, "test_acc": 0.3968, "test_loss": 1.7498909545898438}, {"step": 1000, "epoch": 2, "train_loss": 1.779970407485962, "cos_r_negg": [0.7097252607345581, 0.13704977929592133, 0.13378781080245972, 0.1518288254737854, 0.1883062720298767, 0.09800134599208832, 0.16196227073669434, 0.10695038735866547, 0.15700779855251312, 0.17132630944252014], "cos_innovation_negg": [0.7097252607345581, 0.13704977929592133, 0.13378781080245972, 0.1518288254737854, 0.1883062720298767, 0.09800134599208832, 0.16196227073669434, 0.10695038735866547, 0.15700779855251312, 0.17132630944252014], "cos_apical_negg": [0.7097252607345581, 0.13704977929592133, 0.13378781080245972, 0.1518288254737854, 0.1883062720298767, 0.09800134599208832, 0.16196227073669434, 0.10695038735866547, 0.15700779855251312, 0.17132630944252014], "cos_Ac_negg": [0.7097252607345581, 0.13704977929592133, 0.13378781080245972, 0.1518288254737854, 0.1883062720298767, 0.09800134599208832, 0.16196227073669434, 0.10695038735866547, 0.15700779855251312, 0.17132630944252014], "r_norm": [1.5233681201934814, 1.307344675064087, 1.3632007837295532, 1.4346586465835571, 1.4332027435302734, 1.3592244386672974, 1.3217759132385254, 1.4590411186218262, 1.4288039207458496, 1.3240617513656616]}, {"epoch_end": 2, "step": 1173, "test_acc": 0.3965, "test_loss": 1.7326806274414062}, {"step": 1500, "epoch": 3, "train_loss": 1.6082686185836792, "cos_r_negg": [0.7592759132385254, 0.21454685926437378, 0.23934708535671234, 0.2816928029060364, 0.28407391905784607, 0.21849992871284485, 0.23273241519927979, 0.1718161702156067, 0.24690139293670654, 0.26095688343048096], "cos_innovation_negg": [0.7592759132385254, 0.21454685926437378, 0.23934708535671234, 0.2816928029060364, 0.28407391905784607, 0.21849992871284485, 0.23273241519927979, 0.1718161702156067, 0.24690139293670654, 0.26095688343048096], "cos_apical_negg": [0.7592759132385254, 0.21454685926437378, 0.23934708535671234, 0.2816928029060364, 0.28407391905784607, 0.21849992871284485, 0.23273241519927979, 0.1718161702156067, 0.24690139293670654, 0.26095688343048096], "cos_Ac_negg": [0.7592759132385254, 0.21454685926437378, 0.23934708535671234, 0.2816928029060364, 0.28407391905784607, 0.21849992871284485, 0.23273241519927979, 0.1718161702156067, 0.24690139293670654, 0.26095688343048096], "r_norm": [1.310196876525879, 1.071918249130249, 1.1155903339385986, 1.168304443359375, 1.1752805709838867, 1.1034331321716309, 1.0822980403900146, 1.1760265827178955, 1.1643164157867432, 1.0854392051696777]}, {"epoch_end": 3, "step": 1564, "test_acc": 0.4041, "test_loss": 1.7226017333984376}, {"epoch_end": 4, "step": 1955, "test_acc": 0.41, "test_loss": 1.7017920776367188}], "final": {"test_acc": 0.41, "test_loss": 1.7017920776367188, "wall_s": 15.43833065032959, "cos_r_negg": [0.8203680515289307, 0.3326583206653595, 0.3182148039340973, 0.34564048051834106, 0.40990233421325684, 0.3242473304271698, 0.3651769757270813, 0.31921327114105225, 0.3669641613960266, 0.39996254444122314], "cos_innovation_negg": [0.8203680515289307, 0.3326583206653595, 0.3182148039340973, 0.34564048051834106, 0.40990233421325684, 0.3242473304271698, 0.3651769757270813, 0.31921327114105225, 0.3669641613960266, 0.39996254444122314], "cos_apical_negg": [0.8203680515289307, 0.3326583206653595, 0.3182148039340973, 0.34564048051834106, 0.40990233421325684, 0.3242473304271698, 0.3651769757270813, 0.31921327114105225, 0.3669641613960266, 0.39996254444122314], "cos_Ac_negg": [0.8203680515289307, 0.3326583206653595, 0.3182148039340973, 0.34564048051834106, 0.40990233421325684, 0.3242473304271698, 0.3651769757270813, 0.31921327114105225, 0.3669641613960266, 0.39996254444122314]}}
\ No newline at end of file |
