results: confirm C2 feedback amortization bottleneck

author: YurenHao0426 <Blackhao0426@gmail.com> 2026-07-22 04:51:49 -0500
committer: YurenHao0426 <Blackhao0426@gmail.com> 2026-07-22 04:51:49 -0500
commit: d344f7fa4b8a6b907d364d1900348ca8e4b709ce (patch)
tree: 13fcf933e308c918c8047aa343787a662044e962
parent: e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c (diff)
33 files changed, 96 insertions, 5 deletions
diff --git a/NOVELTY.md b/NOVELTY.md
index 3755595..c044927 100644
--- a/NOVELTY.md
+++ b/NOVELTY.md
@@ -76,11 +76,16 @@ identifies the traffic relationship without observing the teaching component.
 
 - The inherited learned-feedback backbone preserves performance with depth better than fixed DFA
   in the audited flattened-CIFAR MLP setting.
-- Controlled pilots show that per-cell residualization can protect learning from soma-coupled and
-  endogenous top-down traffic; the latter remains a validation-only, single-seed result until C1 is
-  completed.
-- Simultaneous batched perturbations make the inherited calibration practical in this codebase;
-  a hardware-independent query/FLOP advantage is not yet established.
+- Frozen controls show that per-cell residualization protects learning from soma-coupled traffic.
+  It mitigates but does not fully remove endogenous top-down traffic, so arbitrary contextual
+  innovation is not an allowed teaching-signal claim.
+- Simultaneous batched perturbations have a frozen hardware-independent advantage: K1/every4
+  retains 112.9% of the K16/every4 gain over DFA with 16x fewer logical loss queries, 11x less
+  calibration work, and 5.3x less total forward-equivalent work on CIFAR-10 d20/w64.
+- Direct node perturbation solves the frozen useful-depth task while the amortized context
+  vectorizer does not. This identifies feedback amortization as the current engineering
+  bottleneck; it is not a novelty claim and the direct estimator's 68.4x work precludes presenting
+  it as the scalable algorithm.
 
 ## Claims currently forbidden
 
diff --git a/RESULTS.md b/RESULTS.md
index 0661b42..2affc24 100644
--- a/RESULTS.md
+++ b/RESULTS.md
@@ -354,6 +354,36 @@ nonfinite loss and chance accuracy. **The independent recovery gate fails and C2
 prohibited.** A richer vectorizer can improve mean credit without curing coupled A/W basin
 instability; this is now a documented limitation rather than an oral-level scaling result.
 
+## C2 direct node-perturbation causal diagnosis (validation only)
+
+The failed context-vectorizer panel leaves two distinct explanations: the local eligibility/update
+rule may be unable to solve the compositional task, or the learned apical vectorizer may fail to
+amortize a causal direction that would work if it were available. To separate them, direct node
+perturbation replaces every hidden teaching vector by an unamortized antithetic layerwise estimate
+on every update. A task-seed-0 development screen selected `eta=0.03` and K16; the independent
+panel then reused the exact untouched validation splits for task seeds 3--5 and crossed d1/d4 with
+five model seeds. All 30 runs came from clean revision `e8d698e`; test sets remain unevaluated.
+
+| method | depth | validation acc (%) | d1-to-d4 gain | d4 lesion drop |
+|:-------|------:|-------------------:|--------------:|---------------:|
+| BP | 1 | 68.633 ± 5.351 | — | — |
+| BP | 4 | 97.187 ± 1.532 | +28.553 | +34.170 |
+| context-SDIL | 1 | 63.977 ± 8.636 | — | — |
+| context-SDIL | 4 | 79.120 ± 22.765 | +15.143 | +27.150 |
+| direct NP | 1 | 65.703 ± 8.800 | — | — |
+| direct NP | 4 | 96.150 ± 1.719 | +30.447 | +25.743 |
+
+Direct NP recovers `106.6%` of BP's mean depth gain. All 15 paired runs improve with depth, 14/15
+are harmed by the frozen final-block lesion, all final losses are finite, and its d4 endpoint beats
+context-SDIL by `17.030` points and the stronger FA/DFA endpoint by `15.387` points. **The frozen
+causal diagnosis passes: the local eligibility rule and a forward-only causal direction are
+sufficient here; learning a reliable amortized apical vectorizer is the present bottleneck.**
+
+This is a diagnosis and upper-cost baseline, not the proposed scalable method. At d4 it uses a mean
+`82,560,000` per-example scalar loss evaluations and `43,757,037` forward-equivalent examples,
+`68.4x` ordinary training work, with `108.2` seconds mean wall time. The next recovery must retain
+the vectorizer and disclose how much of this query burden it avoids.
+
 ## Paper-faithful non-backprop baselines
 
 `PETITE` in the initial request is treated as **PEPITA**. Implementations and hyperparameters were
@@ -419,6 +449,8 @@ Five-depth scaling: `experiments/figure_scaling_sweep.sh`; canonical local basel
 `experiments/ep_original_sweep.sh`; exact d1 comparison: `experiments/ep_archmatch_sweep.sh`;
 canonical d2 comparison: `experiments/ep_depth2_sweep.sh`; near-parameter comparison:
 `experiments/ep_matched_sweep.sh`.
+C2 direct causal diagnosis: `experiments/c2_nodepert_validation.sh` followed by
+`python experiments/analyze_c2_nodepert_validation.py`.
 Audited aggregate tables: `experiments/analyze_verified.py`. The claim-locked finalizer is
 `experiments/finalize_claims.sh`; it rejects missing/dirty inputs, regenerates both main figures,
 their captions, source-hash manifest, and `results/audited_tables.md`, then runs all smoke checks.
diff --git a/ROADMAP.md b/ROADMAP.md
index 256c8fe..9a6f46f 100644
--- a/ROADMAP.md
+++ b/ROADMAP.md
@@ -148,6 +148,30 @@ secondary conditions passed at the boundary. Context conditioning improves subst
 linear SDIL's 21.0% recovery but does not make causal-feedback learning reliable enough. The
 predeclared mechanism-recovery branch is closed and no C2 test panel will run.
 
+The follow-up causal diagnosis replaces the learned vectorizer with direct layerwise antithetic
+node-perturbation targets on every update. Task seed 0 and model seeds 0--2 were development-only:
+an estimator/LR screen selected `eta=0.03`, and a predeclared smaller-K-within-one-point rule chose
+K16 over K4. The frozen diagnosis then used task seeds 3--5, model seeds 0--4, d1/d4, the same
+training-only validation splits, and no intermediate held-out metrics. It diagnoses an
+amortization bottleneck only if it recovers at least 70% of BP's depth gain, has at least 10/15
+positive paired gains, beats context-SDIL d4 by at least 2 points, has a mean lesion drop of at
+least 2 points with at least 10/15 positive lesions, and has no nonfinite final loss.
+
+**Direct causal diagnosis status: passed on 2026-07-22.** Direct NP recovered `106.6%` of BP's
+depth gain (`+30.447` versus `+28.553` points), reached `96.150 ± 1.719%` at d4, improved in 15/15
+depth pairs, and had positive lesions in 14/15 pairs. It beat context-SDIL d4 by `17.030` points;
+all losses were finite. The result rules out the local eligibility rule and causal direction as
+the primary C2 failure, but costs `68.4x` ordinary forward-equivalent work and is not a scalable
+solution.
+
+One calibration-quality recovery is now allowed on development task seed 0 only. It keeps the
+bounded context vectorizer, `eta=0.03`, `eta_A=0.01`, and every-four-step calibration fixed, and
+compares layerwise K4 against K16 at d1/d4 over model seeds 0--2. Select the higher mean d4
+validation accuracy, choosing K4 if it lies within one point of K16; any nonfinite candidate is
+ineligible. No result from task seeds 3--5 may enter this selection. A selected protocol must be
+frozen before a new independent panel on task seeds 6--8, and that panel must satisfy the original
+C2 gain/comparator/lesion conditions plus at least 10/15 positive depth gains and finite losses.
+
 ### C3. Causal calibration is genuinely amortized
 
 **Status: passed on 2026-07-22.** The frozen five-seed confirmation retained 112.9% of the K16/e4
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t3_s0.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t3_s0.json
new file mode 100644
index 0000000..2bcfbc8
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t3_s0.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 1, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.0, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 3, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 0, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t3_s0", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 3, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "fdedfe8d9534c1d64d0ac2920c7b61b8e1ff98e9b98c326d64f12f07c65dfc18", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.7108338475227356}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.64, "eval_loss": 0.6193019714355469, "wall_s": 18.81295943260193, "val_acc": 0.64, "val_loss": 0.6193019714355469, "cos_q_negg": [0.9055044651031494], "q_norm": [4.024710655212402], "g_norm": [0.006650309544056654], "loss": 0.5963379740715027}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 18.729612112045288, "diagnostics_wall_s": 0.05301308631896973, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 18.729612112045288, "evaluation_wall_s": 0.025459766387939453}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 84480, "calibration_example_loss_evaluations": 21120000, "calibration_forward_equivalent_examples": 14293333.333333803, "training_forward_equivalent_examples": 14933333.333333803, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17310208, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t3_s1.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t3_s1.json
new file mode 100644
index 0000000..e9f8ee2
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t3_s1.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 1, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.0, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 3, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 1, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t3_s1", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 3, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "fdedfe8d9534c1d64d0ac2920c7b61b8e1ff98e9b98c326d64f12f07c65dfc18", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.6921946406364441}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.6175, "eval_loss": 0.5877452392578125, "wall_s": 17.673343181610107, "val_acc": 0.6175, "val_loss": 0.5877452392578125, "cos_q_negg": [0.8777812719345093], "q_norm": [4.180142402648926], "g_norm": [0.006694769486784935], "loss": 0.5775274038314819}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 17.58417844772339, "diagnostics_wall_s": 0.059008121490478516, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 17.58417844772339, "evaluation_wall_s": 0.02555108070373535}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 84480, "calibration_example_loss_evaluations": 21120000, "calibration_forward_equivalent_examples": 14293333.333333803, "training_forward_equivalent_examples": 14933333.333333803, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17310208, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t3_s2.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t3_s2.json
new file mode 100644
index 0000000..35fd78c
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t3_s2.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 1, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.0, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 3, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 2, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t3_s2", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 3, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "fdedfe8d9534c1d64d0ac2920c7b61b8e1ff98e9b98c326d64f12f07c65dfc18", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.7062802314758301}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.613, "eval_loss": 0.6227874755859375, "wall_s": 17.363320350646973, "val_acc": 0.613, "val_loss": 0.6227874755859375, "cos_q_negg": [0.9042591452598572], "q_norm": [4.0276079177856445], "g_norm": [0.006476795766502619], "loss": 0.6033681631088257}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 17.28355884552002, "diagnostics_wall_s": 0.04944753646850586, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 17.28355884552002, "evaluation_wall_s": 0.02567124366760254}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 84480, "calibration_example_loss_evaluations": 21120000, "calibration_forward_equivalent_examples": 14293333.333333803, "training_forward_equivalent_examples": 14933333.333333803, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17310208, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t3_s3.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t3_s3.json
new file mode 100644
index 0000000..0f2f67d
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t3_s3.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 1, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.0, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 3, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 3, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t3_s3", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 3, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "fdedfe8d9534c1d64d0ac2920c7b61b8e1ff98e9b98c326d64f12f07c65dfc18", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.7154970765113831}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.539, "eval_loss": 0.6285514831542969, "wall_s": 18.934081077575684, "val_acc": 0.539, "val_loss": 0.6285514831542969, "cos_q_negg": [0.9114521145820618], "q_norm": [3.44866681098938], "g_norm": [0.005644296295940876], "loss": 0.6245592832565308}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 18.856585264205933, "diagnostics_wall_s": 0.04728126525878906, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 18.856585264205933, "evaluation_wall_s": 0.024502277374267578}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 84480, "calibration_example_loss_evaluations": 21120000, "calibration_forward_equivalent_examples": 14293333.333333803, "training_forward_equivalent_examples": 14933333.333333803, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17310208, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t3_s4.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t3_s4.json
new file mode 100644
index 0000000..468ef39
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t3_s4.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 1, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.0, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 3, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 4, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t3_s4", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 3, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "fdedfe8d9534c1d64d0ac2920c7b61b8e1ff98e9b98c326d64f12f07c65dfc18", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.7816296219825745}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.657, "eval_loss": 0.5029418029785157, "wall_s": 18.07032346725464, "val_acc": 0.657, "val_loss": 0.5029418029785157, "cos_q_negg": [0.8688189387321472], "q_norm": [5.04344367980957], "g_norm": [0.008306273259222507], "loss": 0.5141857266426086}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 17.99002742767334, "diagnostics_wall_s": 0.04944038391113281, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 17.99002742767334, "evaluation_wall_s": 0.026068925857543945}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 84480, "calibration_example_loss_evaluations": 21120000, "calibration_forward_equivalent_examples": 14293333.333333803, "training_forward_equivalent_examples": 14933333.333333803, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17310208, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t4_s0.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t4_s0.json
new file mode 100644
index 0000000..177ddfb
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t4_s0.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 1, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.0, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 4, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 0, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t4_s0", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 4, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "9e82c4f08ab216367fb95592e461929e3aecf2d9041f5e7756f77d140378bd3a", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.6991833448410034}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.636, "eval_loss": 0.6093101501464844, "wall_s": 18.633137941360474, "val_acc": 0.636, "val_loss": 0.6093101501464844, "cos_q_negg": [0.888920783996582], "q_norm": [3.8982162475585938], "g_norm": [0.006465675309300423], "loss": 0.6195128560066223}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 18.54052495956421, "diagnostics_wall_s": 0.061365604400634766, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 18.54052495956421, "evaluation_wall_s": 0.026294708251953125}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 84480, "calibration_example_loss_evaluations": 21120000, "calibration_forward_equivalent_examples": 14293333.333333803, "training_forward_equivalent_examples": 14933333.333333803, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17310208, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t4_s1.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t4_s1.json
new file mode 100644
index 0000000..a6591f3
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t4_s1.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 1, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.0, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 4, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 1, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t4_s1", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 4, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "9e82c4f08ab216367fb95592e461929e3aecf2d9041f5e7756f77d140378bd3a", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.6932581663131714}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.696, "eval_loss": 0.6020486145019531, "wall_s": 17.607300519943237, "val_acc": 0.696, "val_loss": 0.6020486145019531, "cos_q_negg": [0.8928226232528687], "q_norm": [4.070675373077393], "g_norm": [0.0067178476601839066], "loss": 0.614193320274353}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 17.527471780776978, "diagnostics_wall_s": 0.049399614334106445, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 17.527471780776978, "evaluation_wall_s": 0.02521491050720215}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 84480, "calibration_example_loss_evaluations": 21120000, "calibration_forward_equivalent_examples": 14293333.333333803, "training_forward_equivalent_examples": 14933333.333333803, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17310208, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t4_s2.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t4_s2.json
new file mode 100644
index 0000000..f9df5d7
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t4_s2.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 1, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.0, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 4, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 2, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t4_s2", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 4, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "9e82c4f08ab216367fb95592e461929e3aecf2d9041f5e7756f77d140378bd3a", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.6969773173332214}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.5075, "eval_loss": 0.602968505859375, "wall_s": 19.394821405410767, "val_acc": 0.5075, "val_loss": 0.602968505859375, "cos_q_negg": [0.9142959117889404], "q_norm": [4.0681047439575195], "g_norm": [0.0066280728206038475], "loss": 0.612793505191803}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 19.30233669281006, "diagnostics_wall_s": 0.06157112121582031, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 19.30233669281006, "evaluation_wall_s": 0.025992155075073242}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 84480, "calibration_example_loss_evaluations": 21120000, "calibration_forward_equivalent_examples": 14293333.333333803, "training_forward_equivalent_examples": 14933333.333333803, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17310208, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t4_s3.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t4_s3.json
new file mode 100644
index 0000000..0e1d13f
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t4_s3.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 1, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.0, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 4, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 3, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t4_s3", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 4, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "9e82c4f08ab216367fb95592e461929e3aecf2d9041f5e7756f77d140378bd3a", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.701652467250824}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.5785, "eval_loss": 0.60030517578125, "wall_s": 17.77542781829834, "val_acc": 0.5785, "val_loss": 0.60030517578125, "cos_q_negg": [0.9129284620285034], "q_norm": [3.966425657272339], "g_norm": [0.006561105139553547], "loss": 0.6114529967308044}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 17.693135499954224, "diagnostics_wall_s": 0.050975799560546875, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 17.693135499954224, "evaluation_wall_s": 0.025805234909057617}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 84480, "calibration_example_loss_evaluations": 21120000, "calibration_forward_equivalent_examples": 14293333.333333803, "training_forward_equivalent_examples": 14933333.333333803, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17310208, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t4_s4.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t4_s4.json
new file mode 100644
index 0000000..00155cd
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t4_s4.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 1, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.0, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 4, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 4, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t4_s4", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 4, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "9e82c4f08ab216367fb95592e461929e3aecf2d9041f5e7756f77d140378bd3a", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.8244266510009766}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.624, "eval_loss": 0.5605845031738281, "wall_s": 18.517950534820557, "val_acc": 0.624, "val_loss": 0.5605845031738281, "cos_q_negg": [0.8865610957145691], "q_norm": [4.333579063415527], "g_norm": [0.007202482316643], "loss": 0.5755634903907776}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 18.4384822845459, "diagnostics_wall_s": 0.049007415771484375, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 18.4384822845459, "evaluation_wall_s": 0.025658130645751953}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 84480, "calibration_example_loss_evaluations": 21120000, "calibration_forward_equivalent_examples": 14293333.333333803, "training_forward_equivalent_examples": 14933333.333333803, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17310208, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t5_s0.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t5_s0.json
new file mode 100644
index 0000000..849c5b7
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t5_s0.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 1, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.0, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 5, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 0, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t5_s0", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 5, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "2335c66a37fdb438e9b862f9f7047842ab8a843875e1727368e9499ca6b633e8", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.7017548084259033}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.6475, "eval_loss": 0.6102030639648437, "wall_s": 18.990405321121216, "val_acc": 0.6475, "val_loss": 0.6102030639648437, "cos_q_negg": [0.9182716608047485], "q_norm": [4.017208576202393], "g_norm": [0.006603345740586519], "loss": 0.6318932771682739}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 18.896069049835205, "diagnostics_wall_s": 0.06342840194702148, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 18.896069049835205, "evaluation_wall_s": 0.02587294578552246}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 84480, "calibration_example_loss_evaluations": 21120000, "calibration_forward_equivalent_examples": 14293333.333333803, "training_forward_equivalent_examples": 14933333.333333803, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17310208, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t5_s1.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t5_s1.json
new file mode 100644
index 0000000..13432ec
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t5_s1.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 1, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.0, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 5, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 1, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t5_s1", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 5, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "2335c66a37fdb438e9b862f9f7047842ab8a843875e1727368e9499ca6b633e8", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.6929494142532349}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.7625, "eval_loss": 0.5963175659179687, "wall_s": 19.085628747940063, "val_acc": 0.7625, "val_loss": 0.5963175659179687, "cos_q_negg": [0.8914648294448853], "q_norm": [4.0850324630737305], "g_norm": [0.006710999645292759], "loss": 0.6193259954452515}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 18.9988796710968, "diagnostics_wall_s": 0.0557708740234375, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 18.9988796710968, "evaluation_wall_s": 0.02563190460205078}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 84480, "calibration_example_loss_evaluations": 21120000, "calibration_forward_equivalent_examples": 14293333.333333803, "training_forward_equivalent_examples": 14933333.333333803, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17310208, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t5_s2.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t5_s2.json
new file mode 100644
index 0000000..9ee846b
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t5_s2.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 1, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.0, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 5, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 2, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t5_s2", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 5, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "2335c66a37fdb438e9b862f9f7047842ab8a843875e1727368e9499ca6b633e8", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.6991405487060547}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.76, "eval_loss": 0.6001518249511719, "wall_s": 17.768728256225586, "val_acc": 0.76, "val_loss": 0.6001518249511719, "cos_q_negg": [0.9026006460189819], "q_norm": [4.0087127685546875], "g_norm": [0.006459907162934542], "loss": 0.6219915747642517}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 17.68324303627014, "diagnostics_wall_s": 0.05140399932861328, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 17.68324303627014, "evaluation_wall_s": 0.028936147689819336}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 84480, "calibration_example_loss_evaluations": 21120000, "calibration_forward_equivalent_examples": 14293333.333333803, "training_forward_equivalent_examples": 14933333.333333803, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17310208, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t5_s3.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t5_s3.json
new file mode 100644
index 0000000..850c35b
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t5_s3.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 1, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.0, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 5, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 3, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t5_s3", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 5, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "2335c66a37fdb438e9b862f9f7047842ab8a843875e1727368e9499ca6b633e8", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.7045606374740601}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.7515, "eval_loss": 0.6048772888183593, "wall_s": 18.150933504104614, "val_acc": 0.7515, "val_loss": 0.6048772888183593, "cos_q_negg": [0.8935335874557495], "q_norm": [4.084601402282715], "g_norm": [0.00661036791279912], "loss": 0.6259421110153198}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 18.069716691970825, "diagnostics_wall_s": 0.0499269962310791, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 18.069716691970825, "evaluation_wall_s": 0.02631092071533203}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 84480, "calibration_example_loss_evaluations": 21120000, "calibration_forward_equivalent_examples": 14293333.333333803, "training_forward_equivalent_examples": 14933333.333333803, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17310208, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t5_s4.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t5_s4.json
new file mode 100644
index 0000000..a1a79c0
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t5_s4.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 1, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.0, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 5, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 4, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d1_t5_s4", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 5, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "2335c66a37fdb438e9b862f9f7047842ab8a843875e1727368e9499ca6b633e8", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.7986906170845032}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.8255, "eval_loss": 0.5226148986816406, "wall_s": 17.464256286621094, "val_acc": 0.8255, "val_loss": 0.5226148986816406, "cos_q_negg": [0.8782387971878052], "q_norm": [4.889838695526123], "g_norm": [0.008108612149953842], "loss": 0.5434975624084473}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 17.384549617767334, "diagnostics_wall_s": 0.049602508544921875, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 17.384549617767334, "evaluation_wall_s": 0.025488615036010742}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 84480, "calibration_example_loss_evaluations": 21120000, "calibration_forward_equivalent_examples": 14293333.333333803, "training_forward_equivalent_examples": 14933333.333333803, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17310208, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t3_s0.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t3_s0.json
new file mode 100644
index 0000000..3c8b5d9
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t3_s0.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 4, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.3333333333333333, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 3, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 0, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t3_s0", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 3, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "fdedfe8d9534c1d64d0ac2920c7b61b8e1ff98e9b98c326d64f12f07c65dfc18", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.6894387006759644}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.9645, "eval_loss": 0.092102294921875, "wall_s": 104.6530327796936, "val_acc": 0.9645, "val_loss": 0.092102294921875, "cos_q_negg": [0.8243465423583984, 0.8042224645614624, 0.8358355760574341, 0.9020771980285645], "q_norm": [2.5178110599517822, 2.1947124004364014, 1.275694727897644, 0.5596175193786621], "g_norm": [0.004142231307923794, 0.003403439885005355, 0.002164679579436779, 0.0008841060334816575], "loss": 0.08117692172527313, "residual_lesion": {"interior_layers": [1, 2, 3], "lesioned_layers": [3], "branch_to_skip_rms": [0.3122805953025818, 1.654921531677246, 0.9850496053695679], "lesion_eval_acc": 0.6205, "lesion_eval_loss": 0.9441365051269531, "lesion_acc_drop": 0.344}}, "lesion_protocol": {"fraction_of_interior_blocks": 0.3333333333333333, "selection": "final_contiguous_interior_blocks", "probe_source": "training_prefix", "probe_examples": 512, "evaluation_split": "validation"}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 104.48470520973206, "diagnostics_wall_s": 0.10285401344299316, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 104.48470520973206, "evaluation_wall_s": 0.05926394462585449}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 330240, "calibration_example_loss_evaluations": 82560000, "calibration_forward_equivalent_examples": 43117037.03703508, "training_forward_equivalent_examples": 43757037.03703508, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17543168, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t3_s1.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t3_s1.json
new file mode 100644
index 0000000..ddd11f8
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t3_s1.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 4, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.3333333333333333, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 3, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 1, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t3_s1", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 3, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "fdedfe8d9534c1d64d0ac2920c7b61b8e1ff98e9b98c326d64f12f07c65dfc18", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.745452880859375}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.9695, "eval_loss": 0.06932980728149414, "wall_s": 104.64046216011047, "val_acc": 0.9695, "val_loss": 0.06932980728149414, "cos_q_negg": [0.7620583772659302, 0.8035386800765991, 0.8567513227462769, 0.8373085260391235], "q_norm": [3.1926803588867188, 3.0902867317199707, 1.0562466382980347, 0.49712592363357544], "g_norm": [0.005297740921378136, 0.004952254239469767, 0.0016988106071949005, 0.0007933920715004206], "loss": 0.08321386575698853, "residual_lesion": {"interior_layers": [1, 2, 3], "lesioned_layers": [3], "branch_to_skip_rms": [1.078613519668579, 0.2608351707458496, 0.4270806908607483], "lesion_eval_acc": 0.938, "lesion_eval_loss": 0.15026262664794923, "lesion_acc_drop": 0.031500000000000083}}, "lesion_protocol": {"fraction_of_interior_blocks": 0.3333333333333333, "selection": "final_contiguous_interior_blocks", "probe_source": "training_prefix", "probe_examples": 512, "evaluation_split": "validation"}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 104.4727554321289, "diagnostics_wall_s": 0.1027841567993164, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 104.4727554321289, "evaluation_wall_s": 0.05913376808166504}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 330240, "calibration_example_loss_evaluations": 82560000, "calibration_forward_equivalent_examples": 43117037.03703508, "training_forward_equivalent_examples": 43757037.03703508, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17543168, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t3_s2.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t3_s2.json
new file mode 100644
index 0000000..1972425
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t3_s2.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 4, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.3333333333333333, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 3, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 2, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t3_s2", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 3, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "fdedfe8d9534c1d64d0ac2920c7b61b8e1ff98e9b98c326d64f12f07c65dfc18", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.6903389096260071}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.946, "eval_loss": 0.0993888053894043, "wall_s": 104.2924473285675, "val_acc": 0.946, "val_loss": 0.0993888053894043, "cos_q_negg": [0.7600505352020264, 0.6145294308662415, 0.8876035809516907, 0.8736376762390137], "q_norm": [3.8306212425231934, 3.490138053894043, 2.6177978515625, 0.7323014736175537], "g_norm": [0.006034054793417454, 0.0052934568375349045, 0.003980895038694143, 0.0011982785072177649], "loss": 0.11029600352048874, "residual_lesion": {"interior_layers": [1, 2, 3], "lesioned_layers": [3], "branch_to_skip_rms": [0.6309546232223511, 0.48547041416168213, 0.7598876357078552], "lesion_eval_acc": 0.8255, "lesion_eval_loss": 0.9902316589355469, "lesion_acc_drop": 0.12049999999999994}}, "lesion_protocol": {"fraction_of_interior_blocks": 0.3333333333333333, "selection": "final_contiguous_interior_blocks", "probe_source": "training_prefix", "probe_examples": 512, "evaluation_split": "validation"}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 104.11387038230896, "diagnostics_wall_s": 0.10988712310791016, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 104.11387038230896, "evaluation_wall_s": 0.0614924430847168}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 330240, "calibration_example_loss_evaluations": 82560000, "calibration_forward_equivalent_examples": 43117037.03703508, "training_forward_equivalent_examples": 43757037.03703508, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17543168, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t3_s3.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t3_s3.json
new file mode 100644
index 0000000..9b6c280
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t3_s3.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 4, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.3333333333333333, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 3, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 3, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t3_s3", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 3, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "fdedfe8d9534c1d64d0ac2920c7b61b8e1ff98e9b98c326d64f12f07c65dfc18", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.7031912803649902}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.9415, "eval_loss": 0.12221705627441407, "wall_s": 109.67636036872864, "val_acc": 0.9415, "val_loss": 0.12221705627441407, "cos_q_negg": [0.7722295522689819, 0.6674389243125916, 0.9851507544517517, 0.8608128428459167], "q_norm": [3.6546292304992676, 2.7774457931518555, 1.416299819946289, 0.7732417583465576], "g_norm": [0.005459075793623924, 0.0044353981502354145, 0.0023270314559340477, 0.0012665309477597475], "loss": 0.12178049981594086, "residual_lesion": {"interior_layers": [1, 2, 3], "lesioned_layers": [3], "branch_to_skip_rms": [1.4661061763763428, 0.24139204621315002, 0.7864852547645569], "lesion_eval_acc": 0.7115, "lesion_eval_loss": 3.7340391845703125, "lesion_acc_drop": 0.22999999999999998}}, "lesion_protocol": {"fraction_of_interior_blocks": 0.3333333333333333, "selection": "final_contiguous_interior_blocks", "probe_source": "training_prefix", "probe_examples": 512, "evaluation_split": "validation"}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 109.50290536880493, "diagnostics_wall_s": 0.10906124114990234, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 109.50290536880493, "evaluation_wall_s": 0.058573007583618164}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 330240, "calibration_example_loss_evaluations": 82560000, "calibration_forward_equivalent_examples": 43117037.03703508, "training_forward_equivalent_examples": 43757037.03703508, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17543168, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t3_s4.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t3_s4.json
new file mode 100644
index 0000000..02d1aa4
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t3_s4.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 4, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.3333333333333333, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 3, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 4, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t3_s4", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 3, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "fdedfe8d9534c1d64d0ac2920c7b61b8e1ff98e9b98c326d64f12f07c65dfc18", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 1.1787136793136597}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.9845, "eval_loss": 0.04732159805297852, "wall_s": 106.5337553024292, "val_acc": 0.9845, "val_loss": 0.04732159805297852, "cos_q_negg": [0.7479798793792725, 0.7169545292854309, 0.8241156339645386, 0.9454904198646545], "q_norm": [3.865180730819702, 2.200723886489868, 1.356305718421936, 0.5820520520210266], "g_norm": [0.0055169700644910336, 0.003493786323815584, 0.0023313499987125397, 0.0009972628904506564], "loss": 0.06278812885284424, "residual_lesion": {"interior_layers": [1, 2, 3], "lesioned_layers": [3], "branch_to_skip_rms": [0.8642225861549377, 0.43596547842025757, 0.8288291692733765], "lesion_eval_acc": 0.66, "lesion_eval_loss": 2.1322647705078124, "lesion_acc_drop": 0.3245}}, "lesion_protocol": {"fraction_of_interior_blocks": 0.3333333333333333, "selection": "final_contiguous_interior_blocks", "probe_source": "training_prefix", "probe_examples": 512, "evaluation_split": "validation"}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 106.36792159080505, "diagnostics_wall_s": 0.10001778602600098, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 106.36792159080505, "evaluation_wall_s": 0.05910158157348633}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 330240, "calibration_example_loss_evaluations": 82560000, "calibration_forward_equivalent_examples": 43117037.03703508, "training_forward_equivalent_examples": 43757037.03703508, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17543168, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t4_s0.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t4_s0.json
new file mode 100644
index 0000000..adcd93a
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t4_s0.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 4, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.3333333333333333, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 4, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 0, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t4_s0", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 4, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "9e82c4f08ab216367fb95592e461929e3aecf2d9041f5e7756f77d140378bd3a", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.6999248266220093}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.964, "eval_loss": 0.07596994400024414, "wall_s": 109.28996562957764, "val_acc": 0.964, "val_loss": 0.07596994400024414, "cos_q_negg": [0.7704330086708069, 0.8587152361869812, 0.8399463295936584, 0.8090322017669678], "q_norm": [3.0179882049560547, 1.949363112449646, 1.2209587097167969, 0.606120228767395], "g_norm": [0.004720879718661308, 0.0030918363481760025, 0.001858289004303515, 0.0008893102640286088], "loss": 0.0745125412940979, "residual_lesion": {"interior_layers": [1, 2, 3], "lesioned_layers": [3], "branch_to_skip_rms": [0.6667241454124451, 0.6512784957885742, 0.517181396484375], "lesion_eval_acc": 0.718, "lesion_eval_loss": 0.6882769775390625, "lesion_acc_drop": 0.246}}, "lesion_protocol": {"fraction_of_interior_blocks": 0.3333333333333333, "selection": "final_contiguous_interior_blocks", "probe_source": "training_prefix", "probe_examples": 512, "evaluation_split": "validation"}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 109.11446118354797, "diagnostics_wall_s": 0.10937976837158203, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 109.11446118354797, "evaluation_wall_s": 0.059778690338134766}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 330240, "calibration_example_loss_evaluations": 82560000, "calibration_forward_equivalent_examples": 43117037.03703508, "training_forward_equivalent_examples": 43757037.03703508, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17543168, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t4_s1.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t4_s1.json
new file mode 100644
index 0000000..068557d
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t4_s1.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 4, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.3333333333333333, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 4, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 1, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t4_s1", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 4, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "9e82c4f08ab216367fb95592e461929e3aecf2d9041f5e7756f77d140378bd3a", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.7811087965965271}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.9295, "eval_loss": 0.18918651580810547, "wall_s": 107.74593138694763, "val_acc": 0.9295, "val_loss": 0.18918651580810547, "cos_q_negg": [0.6802083253860474, 0.9039915204048157, 0.7022525072097778, 0.8777793645858765], "q_norm": [3.5015430450439453, 2.4238767623901367, 1.5511882305145264, 0.6801866292953491], "g_norm": [0.005997473374009132, 0.004044964909553528, 0.0024728726129978895, 0.0011092524509876966], "loss": 0.12787488102912903, "residual_lesion": {"interior_layers": [1, 2, 3], "lesioned_layers": [3], "branch_to_skip_rms": [0.7836337685585022, 1.5728302001953125, 0.9658351540565491], "lesion_eval_acc": 0.5, "lesion_eval_loss": 7.5058388671875, "lesion_acc_drop": 0.4295}}, "lesion_protocol": {"fraction_of_interior_blocks": 0.3333333333333333, "selection": "final_contiguous_interior_blocks", "probe_source": "training_prefix", "probe_examples": 512, "evaluation_split": "validation"}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 107.54689192771912, "diagnostics_wall_s": 0.11980533599853516, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 107.54689192771912, "evaluation_wall_s": 0.07260680198669434}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 330240, "calibration_example_loss_evaluations": 82560000, "calibration_forward_equivalent_examples": 43117037.03703508, "training_forward_equivalent_examples": 43757037.03703508, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17543168, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t4_s2.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t4_s2.json
new file mode 100644
index 0000000..976a397
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t4_s2.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 4, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.3333333333333333, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 4, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 2, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t4_s2", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 4, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "9e82c4f08ab216367fb95592e461929e3aecf2d9041f5e7756f77d140378bd3a", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.6940909624099731}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.972, "eval_loss": 0.06581241798400879, "wall_s": 109.71423697471619, "val_acc": 0.972, "val_loss": 0.06581241798400879, "cos_q_negg": [0.8650051355361938, 0.695030927658081, 0.9187275171279907, 0.8333762288093567], "q_norm": [2.8076789379119873, 2.0914947986602783, 1.0709638595581055, 0.548862874507904], "g_norm": [0.004339948296546936, 0.0033602560870349407, 0.0019460255280137062, 0.000849980628117919], "loss": 0.07670897990465164, "residual_lesion": {"interior_layers": [1, 2, 3], "lesioned_layers": [3], "branch_to_skip_rms": [1.2442280054092407, 0.41729801893234253, 0.3650530278682709], "lesion_eval_acc": 0.726, "lesion_eval_loss": 1.5216163940429688, "lesion_acc_drop": 0.246}}, "lesion_protocol": {"fraction_of_interior_blocks": 0.3333333333333333, "selection": "final_contiguous_interior_blocks", "probe_source": "training_prefix", "probe_examples": 512, "evaluation_split": "validation"}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 109.54320931434631, "diagnostics_wall_s": 0.10414624214172363, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 109.54320931434631, "evaluation_wall_s": 0.060483455657958984}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 330240, "calibration_example_loss_evaluations": 82560000, "calibration_forward_equivalent_examples": 43117037.03703508, "training_forward_equivalent_examples": 43757037.03703508, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17543168, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t4_s3.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t4_s3.json
new file mode 100644
index 0000000..4cf8337
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t4_s3.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 4, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.3333333333333333, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 4, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 3, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t4_s3", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 4, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "9e82c4f08ab216367fb95592e461929e3aecf2d9041f5e7756f77d140378bd3a", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.6956391334533691}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.974, "eval_loss": 0.0654183120727539, "wall_s": 110.91271162033081, "val_acc": 0.974, "val_loss": 0.0654183120727539, "cos_q_negg": [0.8353679776191711, 0.8371627926826477, 0.9017400145530701, 0.9041670560836792], "q_norm": [2.4505650997161865, 1.6429122686386108, 1.4313431978225708, 0.5017905235290527], "g_norm": [0.0037564653903245926, 0.0023783324286341667, 0.0021665040403604507, 0.0008655724814161658], "loss": 0.0668426901102066, "residual_lesion": {"interior_layers": [1, 2, 3], "lesioned_layers": [3], "branch_to_skip_rms": [1.4339812994003296, 0.15234403312206268, 0.3992784321308136], "lesion_eval_acc": 0.6255, "lesion_eval_loss": 3.4688992919921877, "lesion_acc_drop": 0.34850000000000003}}, "lesion_protocol": {"fraction_of_interior_blocks": 0.3333333333333333, "selection": "final_contiguous_interior_blocks", "probe_source": "training_prefix", "probe_examples": 512, "evaluation_split": "validation"}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 110.7424566745758, "diagnostics_wall_s": 0.10639190673828125, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 110.7424566745758, "evaluation_wall_s": 0.05820918083190918}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 330240, "calibration_example_loss_evaluations": 82560000, "calibration_forward_equivalent_examples": 43117037.03703508, "training_forward_equivalent_examples": 43757037.03703508, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17543168, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t4_s4.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t4_s4.json
new file mode 100644
index 0000000..18aca33
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t4_s4.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 4, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.3333333333333333, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 4, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 4, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t4_s4", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 4, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "9e82c4f08ab216367fb95592e461929e3aecf2d9041f5e7756f77d140378bd3a", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 1.0611846446990967}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.985, "eval_loss": 0.04334678840637207, "wall_s": 118.46734189987183, "val_acc": 0.985, "val_loss": 0.04334678840637207, "cos_q_negg": [0.7741137146949768, 0.9083665013313293, 0.831433892250061, 0.7753627896308899], "q_norm": [1.8249229192733765, 1.4332326650619507, 0.568808913230896, 0.4628162980079651], "g_norm": [0.0031160088256001472, 0.002113133668899536, 0.000968209991697222, 0.0007767939823679626], "loss": 0.04386407509446144, "residual_lesion": {"interior_layers": [1, 2, 3], "lesioned_layers": [3], "branch_to_skip_rms": [0.5448504686355591, 0.4182838499546051, 0.44378355145454407], "lesion_eval_acc": 0.8455, "lesion_eval_loss": 0.7496558837890624, "lesion_acc_drop": 0.13949999999999996}}, "lesion_protocol": {"fraction_of_interior_blocks": 0.3333333333333333, "selection": "final_contiguous_interior_blocks", "probe_source": "training_prefix", "probe_examples": 512, "evaluation_split": "validation"}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 118.28693652153015, "diagnostics_wall_s": 0.11436128616333008, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 118.28693652153015, "evaluation_wall_s": 0.05966472625732422}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 330240, "calibration_example_loss_evaluations": 82560000, "calibration_forward_equivalent_examples": 43117037.03703508, "training_forward_equivalent_examples": 43757037.03703508, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17543168, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t5_s0.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t5_s0.json
new file mode 100644
index 0000000..d5a3e68
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t5_s0.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 4, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.3333333333333333, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 5, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 0, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t5_s0", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 5, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "2335c66a37fdb438e9b862f9f7047842ab8a843875e1727368e9499ca6b633e8", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.6962454319000244}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.947, "eval_loss": 0.11073792266845703, "wall_s": 106.71831345558167, "val_acc": 0.947, "val_loss": 0.11073792266845703, "cos_q_negg": [0.6669064164161682, 0.8607051968574524, 0.8237842917442322, 0.8107596039772034], "q_norm": [3.771477460861206, 3.315885543823242, 1.3189444541931152, 0.7269895672798157], "g_norm": [0.00610768236219883, 0.00547762680798769, 0.002274127211421728, 0.0012592547573149204], "loss": 0.11666371673345566, "residual_lesion": {"interior_layers": [1, 2, 3], "lesioned_layers": [3], "branch_to_skip_rms": [0.7322681546211243, 0.8246210217475891, 0.45490562915802], "lesion_eval_acc": 0.972, "lesion_eval_loss": 0.14385370635986328, "lesion_acc_drop": -0.025000000000000022}}, "lesion_protocol": {"fraction_of_interior_blocks": 0.3333333333333333, "selection": "final_contiguous_interior_blocks", "probe_source": "training_prefix", "probe_examples": 512, "evaluation_split": "validation"}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 106.5535626411438, "diagnostics_wall_s": 0.10158801078796387, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 106.5535626411438, "evaluation_wall_s": 0.05739593505859375}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 330240, "calibration_example_loss_evaluations": 82560000, "calibration_forward_equivalent_examples": 43117037.03703508, "training_forward_equivalent_examples": 43757037.03703508, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17543168, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t5_s1.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t5_s1.json
new file mode 100644
index 0000000..2b5761f
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t5_s1.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 4, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.3333333333333333, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 5, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 1, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t5_s1", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 5, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "2335c66a37fdb438e9b862f9f7047842ab8a843875e1727368e9499ca6b633e8", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.761730432510376}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.972, "eval_loss": 0.055815185546875, "wall_s": 115.48394560813904, "val_acc": 0.972, "val_loss": 0.055815185546875, "cos_q_negg": [0.7121231555938721, 0.6177291870117188, 0.9774138331413269, 0.8462144136428833], "q_norm": [2.592625617980957, 1.9409682750701904, 0.8499121069908142, 0.4079047441482544], "g_norm": [0.004195270128548145, 0.003103911876678467, 0.0014719346072524786, 0.000683639373164624], "loss": 0.05627023056149483, "residual_lesion": {"interior_layers": [1, 2, 3], "lesioned_layers": [3], "branch_to_skip_rms": [1.2742042541503906, 0.583968997001648, 0.863634467124939], "lesion_eval_acc": 0.8555, "lesion_eval_loss": 0.4702296142578125, "lesion_acc_drop": 0.11649999999999994}}, "lesion_protocol": {"fraction_of_interior_blocks": 0.3333333333333333, "selection": "final_contiguous_interior_blocks", "probe_source": "training_prefix", "probe_examples": 512, "evaluation_split": "validation"}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 115.32296514511108, "diagnostics_wall_s": 0.09774971008300781, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 115.32296514511108, "evaluation_wall_s": 0.057344913482666016}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 330240, "calibration_example_loss_evaluations": 82560000, "calibration_forward_equivalent_examples": 43117037.03703508, "training_forward_equivalent_examples": 43757037.03703508, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17543168, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t5_s2.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t5_s2.json
new file mode 100644
index 0000000..7ae81f3
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t5_s2.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 4, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.3333333333333333, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 5, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 2, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t5_s2", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 5, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "2335c66a37fdb438e9b862f9f7047842ab8a843875e1727368e9499ca6b633e8", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.6929397583007812}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.9545, "eval_loss": 0.0936049919128418, "wall_s": 105.41306972503662, "val_acc": 0.9545, "val_loss": 0.0936049919128418, "cos_q_negg": [0.8489158153533936, 0.7658065557479858, 0.9231846332550049, 0.8865824341773987], "q_norm": [3.894467830657959, 3.1707496643066406, 2.2136223316192627, 0.8304644227027893], "g_norm": [0.0063291024416685104, 0.005304348189383745, 0.0035176915116608143, 0.0013621838297694921], "loss": 0.11104155331850052, "residual_lesion": {"interior_layers": [1, 2, 3], "lesioned_layers": [3], "branch_to_skip_rms": [1.0505731105804443, 0.474590003490448, 0.8592894673347473], "lesion_eval_acc": 0.5, "lesion_eval_loss": 6.335175537109375, "lesion_acc_drop": 0.4545}}, "lesion_protocol": {"fraction_of_interior_blocks": 0.3333333333333333, "selection": "final_contiguous_interior_blocks", "probe_source": "training_prefix", "probe_examples": 512, "evaluation_split": "validation"}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 105.2321138381958, "diagnostics_wall_s": 0.11944985389709473, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 105.2321138381958, "evaluation_wall_s": 0.05573010444641113}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 330240, "calibration_example_loss_evaluations": 82560000, "calibration_forward_equivalent_examples": 43117037.03703508, "training_forward_equivalent_examples": 43757037.03703508, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17543168, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t5_s3.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t5_s3.json
new file mode 100644
index 0000000..9781672
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t5_s3.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 4, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.3333333333333333, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 5, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 3, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t5_s3", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 5, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "2335c66a37fdb438e9b862f9f7047842ab8a843875e1727368e9499ca6b633e8", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 0.697464108467102}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.941, "eval_loss": 0.13376251220703125, "wall_s": 101.8844244480133, "val_acc": 0.941, "val_loss": 0.13376251220703125, "cos_q_negg": [0.7924220561981201, 0.7786061763763428, NaN, 0.7871522903442383], "q_norm": [2.963897228240967, 2.3717894554138184, 2.107600688934326, 1.1234626770019531], "g_norm": [0.004968085326254368, 0.0037643045652657747, 0.0037643045652657747, 0.0017023957334458828], "loss": 0.1714652180671692, "residual_lesion": {"interior_layers": [1, 2, 3], "lesioned_layers": [3], "branch_to_skip_rms": [2.162729024887085, 0.0, 0.9796817302703857], "lesion_eval_acc": 0.5, "lesion_eval_loss": 13.5627255859375, "lesion_acc_drop": 0.44099999999999995}}, "lesion_protocol": {"fraction_of_interior_blocks": 0.3333333333333333, "selection": "final_contiguous_interior_blocks", "probe_source": "training_prefix", "probe_examples": 512, "evaluation_split": "validation"}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 101.72127199172974, "diagnostics_wall_s": 0.10000753402709961, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 101.72127199172974, "evaluation_wall_s": 0.05746936798095703}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 330240, "calibration_example_loss_evaluations": 82560000, "calibration_forward_equivalent_examples": 43117037.03703508, "training_forward_equivalent_examples": 43757037.03703508, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17543168, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
diff --git a/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t5_s4.json b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t5_s4.json
new file mode 100644
index 0000000..421a107
--- /dev/null
+++ b/results/c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t5_s4.json
@@ -0,0 +1 @@
+{"args": {"mode": "nodepert", "dataset": "tentmap", "depth": 4, "width": 8, "act": "relu", "residual": 1, "residual_lesion_fraction": 0.3333333333333333, "epochs": 80, "batch_size": 256, "train_examples": 0, "val_examples": 2000, "split_seed": 2027, "eval_split": "validation", "task_seed": 5, "task_train_examples": 10000, "task_test_examples": 5000, "task_levels": 2, "task_n_in": 1, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.03, "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": 1, "pert_ndirs": 16, "pert_mode": "layerwise", "use_residual": 0, "raw_scale_control": "none", "learn_A": 0, "learn_P": 0, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.0, "traffic_seed": 1234, "traffic_mode": "none", "predictor_mode": "diagonal", "vectorizer_mode": "linear", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 4, "device": "cuda", "log_every": 100000, "diagnostics": "alignment", "diagnostics_schedule": "final", "eval_every": 0, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "c2_nodepert_val_v1_tent_l2_w8_nodepert_d4_t5_s4", "n_in": 1, "n_out": 2}, "split": {"dataset": "tentmap", "task_seed": 5, "train_examples": 8000, "test_examples": 5000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 2000, "validation_index_sha256": "2335c66a37fdb438e9b862f9f7047842ab8a843875e1727368e9499ca6b633e8", "validation_class_counts": {"0": 1000, "1": 1000}, "split_from_training_only": true}, "provenance": {"git_commit": "e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c", "git_dirty": false}, "diagnostic_protocol": {"probe_source": "training_prefix", "probe_examples": 512, "schedule": "final"}, "steps": [{"step": 0, "epoch": 0, "train_loss": 1.0589196681976318}, {"epoch_end": 0, "step": 32}, {"epoch_end": 1, "step": 64}, {"epoch_end": 2, "step": 96}, {"epoch_end": 3, "step": 128}, {"epoch_end": 4, "step": 160}, {"epoch_end": 5, "step": 192}, {"epoch_end": 6, "step": 224}, {"epoch_end": 7, "step": 256}, {"epoch_end": 8, "step": 288}, {"epoch_end": 9, "step": 320}, {"epoch_end": 10, "step": 352}, {"epoch_end": 11, "step": 384}, {"epoch_end": 12, "step": 416}, {"epoch_end": 13, "step": 448}, {"epoch_end": 14, "step": 480}, {"epoch_end": 15, "step": 512}, {"epoch_end": 16, "step": 544}, {"epoch_end": 17, "step": 576}, {"epoch_end": 18, "step": 608}, {"epoch_end": 19, "step": 640}, {"epoch_end": 20, "step": 672}, {"epoch_end": 21, "step": 704}, {"epoch_end": 22, "step": 736}, {"epoch_end": 23, "step": 768}, {"epoch_end": 24, "step": 800}, {"epoch_end": 25, "step": 832}, {"epoch_end": 26, "step": 864}, {"epoch_end": 27, "step": 896}, {"epoch_end": 28, "step": 928}, {"epoch_end": 29, "step": 960}, {"epoch_end": 30, "step": 992}, {"epoch_end": 31, "step": 1024}, {"epoch_end": 32, "step": 1056}, {"epoch_end": 33, "step": 1088}, {"epoch_end": 34, "step": 1120}, {"epoch_end": 35, "step": 1152}, {"epoch_end": 36, "step": 1184}, {"epoch_end": 37, "step": 1216}, {"epoch_end": 38, "step": 1248}, {"epoch_end": 39, "step": 1280}, {"epoch_end": 40, "step": 1312}, {"epoch_end": 41, "step": 1344}, {"epoch_end": 42, "step": 1376}, {"epoch_end": 43, "step": 1408}, {"epoch_end": 44, "step": 1440}, {"epoch_end": 45, "step": 1472}, {"epoch_end": 46, "step": 1504}, {"epoch_end": 47, "step": 1536}, {"epoch_end": 48, "step": 1568}, {"epoch_end": 49, "step": 1600}, {"epoch_end": 50, "step": 1632}, {"epoch_end": 51, "step": 1664}, {"epoch_end": 52, "step": 1696}, {"epoch_end": 53, "step": 1728}, {"epoch_end": 54, "step": 1760}, {"epoch_end": 55, "step": 1792}, {"epoch_end": 56, "step": 1824}, {"epoch_end": 57, "step": 1856}, {"epoch_end": 58, "step": 1888}, {"epoch_end": 59, "step": 1920}, {"epoch_end": 60, "step": 1952}, {"epoch_end": 61, "step": 1984}, {"epoch_end": 62, "step": 2016}, {"epoch_end": 63, "step": 2048}, {"epoch_end": 64, "step": 2080}, {"epoch_end": 65, "step": 2112}, {"epoch_end": 66, "step": 2144}, {"epoch_end": 67, "step": 2176}, {"epoch_end": 68, "step": 2208}, {"epoch_end": 69, "step": 2240}, {"epoch_end": 70, "step": 2272}, {"epoch_end": 71, "step": 2304}, {"epoch_end": 72, "step": 2336}, {"epoch_end": 73, "step": 2368}, {"epoch_end": 74, "step": 2400}, {"epoch_end": 75, "step": 2432}, {"epoch_end": 76, "step": 2464}, {"epoch_end": 77, "step": 2496}, {"epoch_end": 78, "step": 2528}, {"epoch_end": 79, "step": 2560}], "final": {"eval_split": "validation", "eval_acc": 0.9775, "eval_loss": 0.06386734580993653, "wall_s": 107.50280904769897, "val_acc": 0.9775, "val_loss": 0.06386734580993653, "cos_q_negg": [0.7626428008079529, 0.6460481882095337, 0.9672103524208069, 0.8802282214164734], "q_norm": [3.15193510055542, 2.2168684005737305, 1.6232645511627197, 0.4975271224975586], "g_norm": [0.004930597264319658, 0.003462580731138587, 0.0026548923924565315, 0.0008970160852186382], "loss": 0.05856582522392273, "residual_lesion": {"interior_layers": [1, 2, 3], "lesioned_layers": [3], "branch_to_skip_rms": [0.7508529424667358, 0.3634081184864044, 0.49904224276542664], "lesion_eval_acc": 0.563, "lesion_eval_loss": 1.2330973510742187, "lesion_acc_drop": 0.4145000000000001}}, "lesion_protocol": {"fraction_of_interior_blocks": 0.3333333333333333, "selection": "final_contiguous_interior_blocks", "probe_source": "training_prefix", "probe_examples": 512, "evaluation_split": "validation"}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 107.33608937263489, "diagnostics_wall_s": 0.10220646858215332, "inline_diagnostics_wall_s": 0.0, "optimizer_wall_s_excluding_inline_diagnostics": 107.33608937263489, "evaluation_wall_s": 0.058580636978149414}, "cost": {"train_steps": 2560, "ordinary_training_forward_examples": 640000, "predictor_warmup_forward_examples": 0, "perturbation_events": 2560, "calibration_batch_loss_evaluations": 330240, "calibration_example_loss_evaluations": 82560000, "calibration_forward_equivalent_examples": 43117037.03703508, "training_forward_equivalent_examples": 43757037.03703508, "max_perturbation_batch_expansion": 1}, "hardware": {"device": "cuda", "torch_version": "2.3.1+cu118", "cuda_device_name": "NVIDIA GeForce GTX 1080", "cuda_visible_devices": "7", "device_total_memory_bytes": 8507949056, "peak_memory_allocated_bytes": 17543168, "peak_memory_reserved_bytes": 23068672}}
+\ No newline at end of file
author	YurenHao0426 <Blackhao0426@gmail.com>	2026-07-22 04:51:49 -0500
committer	YurenHao0426 <Blackhao0426@gmail.com>	2026-07-22 04:51:49 -0500
commit	d344f7fa4b8a6b907d364d1900348ca8e4b709ce (patch)
tree	13fcf933e308c918c8047aa343787a662044e962
parent	e8d698e9da1e2441d1d137eb8f8ab6cd46b4282c (diff)