update APPLIED run index

applied/_index/APPLIED_RUNS_INDEX.json

@@ -461,139 +461,131 @@
     }
   },
   {
-    "key": "opt07_sfa_integer_train/storm_reg_0",
-    "tag": "storm_reg_0",
-    "script": "opt07_sfa_integer_train.py",
-    "hf_folder": "applied/opt07_sfa_integer_train",
-    "phase": 5,
+    "key": "opt02_seenn_bayes/default",
+    "tag": "default",
+    "script": "opt02_seenn_bayes.py",
+    "hf_folder": "applied/opt02_seenn_bayes",
+    "phase": 1,
     "status": "ok",
-    "duration_s": 620.6,
+    "duration_s": 586.1,
     "optimization_summary": {
-      "optimization": "sfa_integer_train",
-      "research_section": "3. Training-Time Approximation for Fewer Timesteps (SFA) / spike-storm suppression",
-      "reference": "Yao et al., Scaling Spike-driven Transformer with Efficient Spike Firing Approximation Training, arXiv:2411.16061",
-      "script": "opt07_sfa_integer_train.py",
+      "optimization": "seenn_early_exit_bayesian_fusion",
+      "research_section": "2. Temporal-dimension reduction",
+      "reference": "SEENN: Towards Temporal Spiking Early-Exit Neural Networks, arXiv:2304.01230 (NeurIPS 2023); BayesianSpikeFusion, Front. Neurosci. 2024 (PMC11330889)",
+      "script": "opt02_seenn_bayes.py",
       "baseline_mode": false,
       "method": "harmonic_matched",
       "accuracy": {
-        "test_acc": 0.93,
-        "test_target_acc": 0.93,
-        "best_val_acc": 0.9335714285714286
+        "test_acc": 0.9278571428571428,
+        "test_target_acc": 0.9278571428571428,
+        "best_val_acc": 0.9307142857142857
       },
       "efficiency": {
         "spikes_per_inference": null,
-        "syn_ops_per_inference": 5285655.101461039,
-        "dense_macs_per_inference": 122675200.0,
-        "event_fraction_of_dense": 0.04308658230401123,
-        "syn_ops_per_inference_dense_fanout": 66862272.88798702,
+        "syn_ops_per_inference": 3638362.4326298703,
+        "dense_macs_per_inference": 123955200.0,
+        "event_fraction_of_dense": 0.029352237200455246,
+        "syn_ops_per_inference_dense_fanout": 65214980.21915584,
         "weight_sparsity": 0.7324707846410685,
-        "effective_ticks": 4.083571428571428,
+        "effective_ticks": 7.985,
         "num_ticks": 100,
         "membrane_bytes_per_inference": 451200,
         "weight_bytes": 4907008,
         "input_raster_bytes": null,
-        "train_ms_per_step": 104.8419256499983,
-        "infer_ms_per_sample": 0.6984454099999482,
-        "peak_vram_mb": 2830.736384
+        "train_ms_per_step": 174.30670261383057,
+        "infer_ms_per_sample": 0.6237456846476799,
+        "peak_vram_mb": 10914.396672,
+        "syn_ops_per_inference_early_exit": 1846599.2897727273
       },
       "optimization_specific": {
-        "sfa_levels_D": 4,
-        "soft_reset": true,
-        "storm_reg": 0.0,
-        "trained_T": 100,
-        "infer_mode": "graded",
-        "min_ticks_within_1pct_of_full": 100,
-        "speedup_at_min_ticks": 1.0,
-        "full_T_acc": 0.93,
-        "t_sweep_rows": [
+        "seenn_enabled": true,
+        "bayes_fusion_enabled": true,
+        "conf_threshold": 0.9,
+        "conf_temp": 2.0,
+        "anytime_fracs": "0.25,0.5,1.0",
+        "anytime_weights": "0.3,0.3,1.0",
+        "ic_loss_weight": 0.3,
+        "ic_layer": 0,
+        "mean_exit_tick": 7.985,
+        "mean_exit_ms": 399.25000000000006,
+        "ticks_saved_frac": 0.92015,
+        "exit_tick_histogram": {
+          "counts": [
+            1040,
+            316,
+            39,
+            4,
+            1,
+            0,
+            0,
+            0,
+            0,
+            0
+          ],
+          "edges": [
+            0.0,
+            10.0,
+            20.0,
+            30.0,
+            40.0,
+            50.0,
+            60.0,
+            70.0,
+            80.0,
+            90.0,
+            100.0
+          ]
+        },
+        "acc_fc_only": 0.9278571428571428,
+        "acc_ic_only": 0.935,
+        "acc_fused": 0.9378571428571428,
+        "acc_early_exit": 0.3578571428571429,
+        "syn_ops_saved_frac": 0.49246417200994075,
+        "syn_per_inference_early_exit": 1846599.2897727273,
+        "anytime_acc": {
+          "0.25": 0.6228571428571429,
+          "0.5": 0.8078571428571428,
+          "0.75": 0.8885714285714286,
+          "1.0": 0.9278571428571428
+        },
+        "pareto": [
           {
-            "T": 4,
-            "acc": 0.12142857142857143,
-            "syn_per_inference": 2814512.5584415584,
-            "dense_per_inference": 4907008.0,
-            "infer_mode": "graded",
-            "ticks_speedup_vs_trained_T": 25.0
+            "conf": 0.5,
+            "mean_ticks": 4.405,
+            "acc": 0.22785714285714287,
+            "syn_ops": 1715995.05762987
           },
           {
-            "T": 8,
-            "acc": 0.2307142857142857,
-            "syn_per_inference": 3556388.2735389615,
-            "dense_per_inference": 9814016.0,
-            "infer_mode": "graded",
-            "ticks_speedup_vs_trained_T": 12.5
+            "conf": 0.7,
+            "mean_ticks": 5.097857142857142,
+            "acc": 0.25285714285714284,
+            "syn_ops": 1744549.8368506492
           },
           {
-            "T": 16,
-            "acc": 0.5378571428571428,
-            "syn_per_inference": 4167199.856331169,
-            "dense_per_inference": 19628032.0,
-            "infer_mode": "graded",
-            "ticks_speedup_vs_trained_T": 6.25
+            "conf": 0.8,
+            "mean_ticks": 5.984285714285714,
+            "acc": 0.28285714285714286,
+            "syn_ops": 1779479.5024350649
           },
           {
-            "T": 25,
-            "acc": 0.7707142857142857,
-            "syn_per_inference": 4437607.62012987,
-            "dense_per_inference": 30668800.0,
-            "infer_mode": "graded",
-            "ticks_speedup_vs_trained_T": 4.0
+            "conf": 0.9,
+            "mean_ticks": 7.984285714285714,
+            "acc": 0.3578571428571429,
+            "syn_ops": 1846555.8522727273
           },
           {
-            "T": 50,
-            "acc": 0.8842857142857142,
-            "syn_per_inference": 4779261.4342532465,
-            "dense_per_inference": 61337600.0,
-            "infer_mode": "graded",
-            "ticks_speedup_vs_trained_T": 2.0
+            "conf": 0.95,
+            "mean_ticks": 10.01142857142857,
+            "acc": 0.42857142857142855,
+            "syn_ops": 1902723.0324675327
           },
           {
-            "T": 100,
-            "acc": 0.93,
-            "syn_per_inference": 5285655.101461039,
-            "dense_per_inference": 122675200.0,
-            "infer_mode": "graded",
-            "ticks_speedup_vs_trained_T": 1.0
+            "conf": 0.99,
+            "mean_ticks": 14.458571428571428,
+            "acc": 0.5621428571428572,
+            "syn_ops": 2004484.2800324673
           }
-        ],
-        "storm_ratio_peak_over_mean": 4.048530718036778,
-        "tick_activity_cv": 0.5872231855348422,
-        "mean_integer_activation": 0.10196488050743936,
-        "frac_activations_gt1": 0.02114660758525133,
-        "max_activation": 4.0,
-        "per_layer_activity": [
-          {
-            "layer": 0,
-            "mean_integer_activation": 0.18338739359751344,
-            "peak_tick_activation": 0.6446411237120628,
-            "storm_ratio": 3.515187773085857,
-            "tick_activity_cv": 0.5872231855348422,
-            "frac_activations_gt1": 0.02114660758525133,
-            "max_activation": 4.0
-          },
-          {
-            "layer": 1,
-            "mean_integer_activation": 0.04924224853515625,
-            "peak_tick_activation": 0.12945556640625,
-            "storm_ratio": 2.6289531907509436,
-            "tick_activity_cv": 0.3385586560128827,
-            "frac_activations_gt1": 0.0011788940100814216,
-            "max_activation": 4.0
-          },
-          {
-            "layer": 2,
-            "mean_integer_activation": 0.07326499938964844,
-            "peak_tick_activation": 0.2966156005859375,
-            "storm_ratio": 4.048530718036778,
-            "tick_activity_cv": 0.5675882158492334,
-            "frac_activations_gt1": 0.0014440917584579438,
-            "max_activation": 4.0
-          }
-        ],
-        "expanded_vs_graded_agreement": 1.0,
-        "expanded_acc": 0.8984375,
-        "graded_acc": 0.8984375,
-        "syn_ratio_expanded_over_graded": 1.000382822740147,
-        "expansion_caveat": "exact only at beta=1 (IAF); with beta<1 the expanded mode applies decay per sub-tick and drifts"
+        ]
       },
       "config": {
         "method": "harmonic_matched",
@@ -633,159 +625,94 @@
         "decision_margin": 2.0,
         "export": false,
         "seed": 1234,
-        "hub_dir": "/kaggle/working/hub/opt07_sfa_integer_train/storm_reg_0",
-        "repo_id": "UWU-R-13/SSVEP-SNN",
-        "push_hf": false,
-        "keep_epoch_weights": 1,
-        "no_resume": true,
-        "selftest": false,
         "baseline": false,
         "bench": true,
         "bench_warmup": 10,
         "bench_iters": 50,
         "bench_train_steps": 20,
-        "sfa_levels": 4,
-        "soft_reset": true,
-        "storm_reg": 0.0,
-        "infer_mode": "graded",
-        "t_sweep": "4,8,16,25,50,100",
+        "anytime_fracs": "0.25,0.5,1.0",
+        "anytime_weights": "0.3,0.3,1.0",
+        "ic_layer": 0,
+        "ic_loss_weight": 0.3,
+        "bayes_fusion": true,
+        "fusion_prior_weight": 1.0,
+        "conf_threshold": 0.9,
+        "conf_temp": 2.0,
+        "conf_sweep": "0.5,0.7,0.8,0.9,0.95,0.99",
+        "hub_dir": "/kaggle/working/hub/opt02_seenn_bayes/default",
+        "repo_id": "UWU-R-13/SSVEP-SNN",
+        "push_hf": false,
+        "keep_epoch_weights": 1,
+        "no_resume": true,
+        "selftest": false,
         "sampling_rate": 250.0
       }
     },
     "zip": {
-      "name": "storm_reg_0.zip",
+      "name": "default.zip",
       "files": 81,
-      "bytes": 12259837,
-      "MB": 12.3,
-      "sha256": "194904e94dac6ff6cfd7ba1ce7a3b3ffeb09cd84de849835a09aaac3268a4a08"
+      "bytes": 12636571,
+      "MB": 12.6,
+      "sha256": "1c64c6704af63b5eec37b2f1d2e824ba1c5c3c28328c5b1c03089bd12b4cc747"
     }
   },
   {
-    "key": "opt08_mint_quant/u_bits_0",
-    "tag": "u_bits_0",
-    "script": "opt08_mint_quant.py",
-    "hf_folder": "applied/opt08_mint_quant",
-    "phase": 5,
+    "key": "opt03_ftbc_threshold_balance/control_baseline",
+    "tag": "control_baseline",
+    "script": "opt03_ftbc_threshold_balance.py",
+    "hf_folder": "applied/opt03_ftbc_threshold_balance",
+    "phase": 1,
     "status": "ok",
-    "duration_s": 341.5,
+    "duration_s": 321.7,
     "optimization_summary": {
-      "optimization": "mint_quant",
-      "research_section": "4. Membrane-Potential Memory-Wall Optimizations (multiplier-less integer quantization)",
-      "reference": "Yin et al., MINT: Multiplier-less INTeger Quantization for Energy Efficient Spiking Neural Networks, arXiv:2305.09850 (ASP-DAC 2024)",
-      "script": "opt08_mint_quant.py",
-      "baseline_mode": false,
+      "optimization": "ftbc_threshold_balance",
+      "research_section": "2. Temporal-Dimension Reduction (threshold balancing + FTBC)",
+      "reference": "Rueckauer et al. arXiv:1612.04052; Sengupta et al. arXiv:1802.02627; FTBC: Forward Temporal Bias Correction, ECCV 2024",
+      "script": "opt03_ftbc_threshold_balance.py",
+      "baseline_mode": true,
       "method": "harmonic_matched",
       "accuracy": {
-        "test_acc": 0.9407142857142857,
-        "test_target_acc": 0.9407142857142857,
-        "best_val_acc": 0.9421428571428572
+        "test_acc": 0.9421428571428572,
+        "test_target_acc": 0.9421428571428572,
+        "best_val_acc": 0.945
       },
       "efficiency": {
         "spikes_per_inference": null,
-        "syn_ops_per_inference": 2152696.525466721,
+        "syn_ops_per_inference": 3271536.9975649347,
         "dense_macs_per_inference": 122675200.0,
-        "event_fraction_of_dense": 0.0175479357316452,
-        "syn_ops_per_inference_dense_fanout": 64922516.865259744,
-        "weight_sparsity": 0.8335865765859767,
-        "effective_ticks": 12.575,
+        "event_fraction_of_dense": 0.026668283382174512,
+        "syn_ops_per_inference_dense_fanout": 64848154.78409091,
+        "weight_sparsity": 0.7324707846410685,
+        "effective_ticks": 12.079285714285714,
         "num_ticks": 100,
         "membrane_bytes_per_inference": 451200,
-        "weight_bytes": 613376,
+        "weight_bytes": 4907008,
         "input_raster_bytes": null,
-        "train_ms_per_step": 99.00966735000338,
-        "infer_ms_per_sample": 0.6622065121874954,
-        "peak_vram_mb": 2833.092608
+        "train_ms_per_step": 94.71398885000326,
+        "infer_ms_per_sample": 0.6264105418749466,
+        "peak_vram_mb": 2828.282368
       },
       "optimization_specific": {
-        "w_bits": 4,
-        "u_bits": 0,
-        "leak_shift": 0,
-        "beta_eff": 1.0,
-        "qat": true,
-        "batchnorm": true,
-        "membrane_memory_reduction_x": 1.0,
-        "weight_memory_reduction_x": 8.0,
-        "membrane_bytes_per_inference": 451200.0,
-        "working_set_bytes": 617888.0,
-        "baseline_working_set_bytes": 4911520.0,
-        "l2_cache_fits": true,
-        "baseline_l2_cache_fits": false,
-        "l2_budget_bytes": 2097152,
-        "membrane_saturation_frac": null,
-        "acc_vs_ubits": {
-          "2": 0.8964285714285715,
-          "4": 0.9521428571428572,
-          "8": 0.9521428571428572,
-          "0": 0.9407142857142857
-        },
-        "bits_sweep_rows": [
-          {
-            "u_bits": 2,
-            "acc": 0.8964285714285715,
-            "target_acc": 0.8964285714285715,
-            "membrane_bytes_per_neuron_per_tick": 0.25,
-            "membrane_memory_reduction_x": 16.0,
-            "membrane_saturation_frac": 0.09132064494680851,
-            "acc_drop_vs_fp32": 0.04428571428571426
-          },
-          {
-            "u_bits": 4,
-            "acc": 0.9521428571428572,
-            "target_acc": 0.9521428571428572,
-            "membrane_bytes_per_neuron_per_tick": 0.5,
-            "membrane_memory_reduction_x": 8.0,
-            "membrane_saturation_frac": 0.2783921348625886,
-            "acc_drop_vs_fp32": -0.011428571428571455
-          },
-          {
-            "u_bits": 8,
-            "acc": 0.9521428571428572,
-            "target_acc": 0.9521428571428572,
-            "membrane_bytes_per_neuron_per_tick": 1.0,
-            "membrane_memory_reduction_x": 4.0,
-            "membrane_saturation_frac": 0.32464220412234046,
-            "acc_drop_vs_fp32": -0.011428571428571455
-          },
-          {
-            "u_bits": 0,
-            "acc": 0.9407142857142857,
-            "target_acc": 0.9407142857142857,
-            "membrane_bytes_per_neuron_per_tick": 4.0,
-            "membrane_memory_reduction_x": 1.0,
-            "membrane_saturation_frac": null,
-            "acc_drop_vs_fp32": 0.0
-          }
-        ],
-        "fp32_membrane_acc": 0.9407142857142857,
-        "per_layer_scales": [
-          {
-            "layer": 0,
-            "weight_scale": 0.034521830933434625,
-            "membrane_scale": 1.0,
-            "shared_scale_gauge_factor": 28.967177376200326
-          },
-          {
-            "layer": 1,
-            "weight_scale": 0.02071961760520935,
-            "membrane_scale": 1.0,
-            "shared_scale_gauge_factor": 48.263438981063956
-          },
-          {
-            "layer": 2,
-            "weight_scale": 0.013957016170024872,
-            "membrane_scale": 1.0,
-            "shared_scale_gauge_factor": 71.64855208433981
-          },
-          {
-            "layer": 3,
-            "weight_scale": 0.0242827194077628,
-            "membrane_scale": 1.0,
-            "shared_scale_gauge_factor": 41.18154903524997
-          }
+        "calibrate_thresh": false,
+        "ftbc": false,
+        "calib_percentile": 99.9,
+        "calib_batches": 8,
+        "ftbc_target_ticks": 8,
+        "ftbc_ref_ticks": 100,
+        "trained_T": 100,
+        "thresh_before": [
+          1.0,
+          1.0,
+          1.0,
+          1.0
         ],
-        "shared_scale_note": "weight grid and membrane grid are the same grid up to a per-layer positive scalar that BatchNorm absorbs exactly; exact only with --batchnorm on",
-        "multiplier_less_compare": false,
-        "wall_clock_caveat": "this reference kernel computes in fp32 registers, so the traffic win does not appear in train_ms_per_step; a fixed-point core is where it becomes energy"
+        "thresh_after": null,
+        "min_ticks_within_1pct_of_full": null,
+        "min_ticks_within_1pct_per_arm": null,
+        "speedup_at_min_ticks": null,
+        "full_T_acc": null,
+        "t_sweep_rows": null,
+        "calibration_reference": "same network at full T (NOT a source ANN)"
       },
       "config": {
         "method": "harmonic_matched",
@@ -825,160 +752,296 @@
         "decision_margin": 2.0,
         "export": false,
         "seed": 1234,
-        "hub_dir": "/kaggle/working/hub/opt08_mint_quant/u_bits_0",
+        "hub_dir": "/kaggle/working/hub/opt03_ftbc_threshold_balance/control_baseline",
         "repo_id": "UWU-R-13/SSVEP-SNN",
         "push_hf": false,
         "keep_epoch_weights": 1,
         "no_resume": true,
         "selftest": false,
-        "baseline": false,
+        "baseline": true,
         "bench": true,
         "bench_warmup": 10,
         "bench_iters": 50,
         "bench_train_steps": 20,
-        "w_bits": 4,
-        "u_bits": 0,
-        "leak_shift": 0,
-        "qat": true,
-        "bits_sweep": "2,4,8,0",
-        "l2_bytes": 2097152,
+        "calibrate_thresh": true,
+        "calib_percentile": 99.9,
+        "calib_batches": 8,
+        "ftbc": true,
+        "ftbc_target_ticks": 8,
+        "ftbc_ref_ticks": 0,
+        "ftbc_lr": 1.0,
+        "t_sweep": "4,8,16,25,50,100",
         "sampling_rate": 250.0
       }
     },
     "zip": {
-      "name": "u_bits_0.zip",
-      "files": 81,
-      "bytes": 12238321,
+      "name": "control_baseline.zip",
+      "files": 80,
+      "bytes": 12244248,
       "MB": 12.2,
-      "sha256": "de8d0ec3134b59a7b9e26917542d60cf181033520d851a5638465f805487afa4"
+      "sha256": "c24f218a83b9e351cc6d43ef1b51d7b3f9e3bfa86c64ae22f40bc90ce4f087a4"
     }
   },
   {
-    "key": "opt08_mint_quant/u_bits_3",
-    "tag": "u_bits_3",
-    "script": "opt08_mint_quant.py",
-    "hf_folder": "applied/opt08_mint_quant",
-    "phase": 5,
+    "key": "opt03_ftbc_threshold_balance/default",
+    "tag": "default",
+    "script": "opt03_ftbc_threshold_balance.py",
+    "hf_folder": "applied/opt03_ftbc_threshold_balance",
+    "phase": 1,
     "status": "ok",
-    "duration_s": 341.6,
+    "duration_s": 372.1,
     "optimization_summary": {
-      "optimization": "mint_quant",
-      "research_section": "4. Membrane-Potential Memory-Wall Optimizations (multiplier-less integer quantization)",
-      "reference": "Yin et al., MINT: Multiplier-less INTeger Quantization for Energy Efficient Spiking Neural Networks, arXiv:2305.09850 (ASP-DAC 2024)",
-      "script": "opt08_mint_quant.py",
+      "optimization": "ftbc_threshold_balance",
+      "research_section": "2. Temporal-Dimension Reduction (threshold balancing + FTBC)",
+      "reference": "Rueckauer et al. arXiv:1612.04052; Sengupta et al. arXiv:1802.02627; FTBC: Forward Temporal Bias Correction, ECCV 2024",
+      "script": "opt03_ftbc_threshold_balance.py",
       "baseline_mode": false,
       "method": "harmonic_matched",
       "accuracy": {
-        "test_acc": 0.9578571428571429,
-        "test_target_acc": 0.9578571428571429,
-        "best_val_acc": 0.9628571428571429
+        "test_acc": 0.9292857142857143,
+        "test_target_acc": 0.9292857142857143,
+        "best_val_acc": 0.945
       },
       "efficiency": {
         "spikes_per_inference": null,
-        "syn_ops_per_inference": 1648163.6732954546,
+        "syn_ops_per_inference": 3270969.221590909,
         "dense_macs_per_inference": 122675200.0,
-        "event_fraction_of_dense": 0.013435182280489085,
-        "syn_ops_per_inference_dense_fanout": 64593181.90016234,
-        "weight_sparsity": 0.8624946199394825,
-        "effective_ticks": 11.785714285714286,
+        "event_fraction_of_dense": 0.026663655095658365,
+        "syn_ops_per_inference_dense_fanout": 64847587.008116886,
+        "weight_sparsity": 0.7324707846410685,
+        "effective_ticks": 12.059285714285714,
         "num_ticks": 100,
-        "membrane_bytes_per_inference": 0,
-        "weight_bytes": 613376,
+        "membrane_bytes_per_inference": 451200,
+        "weight_bytes": 4907008,
         "input_raster_bytes": null,
-        "train_ms_per_step": 98.38904980000507,
-        "infer_ms_per_sample": 0.6557833990625284,
-        "peak_vram_mb": 2840.912896
+        "train_ms_per_step": 92.7018213500105,
+        "infer_ms_per_sample": 0.6145065031248009,
+        "peak_vram_mb": 2825.316864
       },
       "optimization_specific": {
-        "w_bits": 4,
-        "u_bits": 3,
-        "leak_shift": 0,
-        "beta_eff": 1.0,
-        "qat": true,
-        "batchnorm": true,
-        "membrane_memory_reduction_x": 10.666666666666666,
-        "weight_memory_reduction_x": 8.0,
-        "membrane_bytes_per_inference": 42300.0,
-        "working_set_bytes": 613799.0,
-        "baseline_working_set_bytes": 4911520.0,
-        "l2_cache_fits": true,
-        "baseline_l2_cache_fits": false,
-        "l2_budget_bytes": 2097152,
-        "membrane_saturation_frac": 0.14972666638962767,
-        "acc_vs_ubits": {
-          "2": 0.5942857142857143,
-          "4": 0.9564285714285714,
-          "8": 0.955,
-          "0": 0.9528571428571428
+        "calibrate_thresh": true,
+        "ftbc": true,
+        "calib_percentile": 99.9,
+        "calib_batches": 8,
+        "ftbc_target_ticks": 8,
+        "ftbc_ref_ticks": 100,
+        "trained_T": 100,
+        "thresh_before": [
+          1.0,
+          1.0,
+          1.0,
+          1.0
+        ],
+        "thresh_after": [
+          2.353323459625244,
+          0.8474444150924683,
+          1.1090540885925293,
+          2.6859006881713867
+        ],
+        "min_ticks_within_1pct_of_full": null,
+        "min_ticks_within_1pct_per_arm": {
+          "raw": 50,
+          "thresh": 50,
+          "ftbc": 100,
+          "both": null
         },
-        "bits_sweep_rows": [
+        "speedup_at_min_ticks": null,
+        "full_T_acc": 0.9442857142857143,
+        "t_sweep_rows": [
+          {
+            "T": 4,
+            "arm": "raw",
+            "acc": 0.2621428571428571,
+            "syn_per_inference": 1869022.726461039,
+            "dense_per_inference": 4907008.0,
+            "event_fraction_of_dense": 0.38088846124991826
+          },
+          {
+            "T": 4,
+            "arm": "thresh",
+            "acc": 0.40214285714285714,
+            "syn_per_inference": 1828094.6006493508,
+            "dense_per_inference": 4907008.0,
+            "event_fraction_of_dense": 0.37254771148719357
+          },
+          {
+            "T": 4,
+            "arm": "ftbc",
+            "acc": 0.18285714285714286,
+            "syn_per_inference": 1982641.3717532465,
+            "dense_per_inference": 4907008.0,
+            "event_fraction_of_dense": 0.4040428244162729
+          },
           {
-            "u_bits": 2,
+            "T": 4,
+            "arm": "both",
+            "acc": 0.27714285714285714,
+            "syn_per_inference": 1953169.6939935065,
+            "dense_per_inference": 4907008.0,
+            "event_fraction_of_dense": 0.3980367861624653
+          },
+          {
+            "T": 8,
+            "arm": "raw",
             "acc": 0.5942857142857143,
-            "target_acc": 0.5942857142857143,
-            "membrane_bytes_per_neuron_per_tick": 0.25,
-            "membrane_memory_reduction_x": 16.0,
-            "membrane_saturation_frac": 0.019098722850177306,
-            "acc_drop_vs_fp32": 0.35857142857142854
+            "syn_per_inference": 2050550.6517857143,
+            "dense_per_inference": 9814016.0,
+            "event_fraction_of_dense": 0.20894103410731288
           },
           {
-            "u_bits": 4,
-            "acc": 0.9564285714285714,
-            "target_acc": 0.9564285714285714,
-            "membrane_bytes_per_neuron_per_tick": 0.5,
-            "membrane_memory_reduction_x": 8.0,
-            "membrane_saturation_frac": 0.20831449468085106,
-            "acc_drop_vs_fp32": -0.0035714285714285587
+            "T": 8,
+            "arm": "thresh",
+            "acc": 0.7785714285714286,
+            "syn_per_inference": 1952516.9748376624,
+            "dense_per_inference": 9814016.0,
+            "event_fraction_of_dense": 0.19895188420700174
           },
           {
-            "u_bits": 8,
-            "acc": 0.955,
-            "target_acc": 0.955,
-            "membrane_bytes_per_neuron_per_tick": 1.0,
-            "membrane_memory_reduction_x": 4.0,
-            "membrane_saturation_frac": 0.2674735427748227,
-            "acc_drop_vs_fp32": -0.002142857142857113
+            "T": 8,
+            "arm": "ftbc",
+            "acc": 0.375,
+            "syn_per_inference": 2353354.2873376627,
+            "dense_per_inference": 9814016.0,
+            "event_fraction_of_dense": 0.2397952364595353
           },
           {
-            "u_bits": 0,
-            "acc": 0.9528571428571428,
-            "target_acc": 0.9528571428571428,
-            "membrane_bytes_per_neuron_per_tick": 4.0,
-            "membrane_memory_reduction_x": 1.0,
-            "membrane_saturation_frac": null,
-            "acc_drop_vs_fp32": 0.0
-          }
-        ],
-        "fp32_membrane_acc": 0.9528571428571428,
-        "per_layer_scales": [
+            "T": 8,
+            "arm": "both",
+            "acc": 0.53,
+            "syn_per_inference": 2284980.627435065,
+            "dense_per_inference": 9814016.0,
+            "event_fraction_of_dense": 0.23282829653375997
+          },
           {
-            "layer": 0,
-            "weight_scale": 0.03138868297849383,
-            "membrane_scale": 0.3333333333333333,
-            "shared_scale_gauge_factor": 10.619538690480226
+            "T": 16,
+            "arm": "raw",
+            "acc": 0.8028571428571428,
+            "syn_per_inference": 2341053.414772727,
+            "dense_per_inference": 19628032.0,
+            "event_fraction_of_dense": 0.11927091899853878
           },
           {
-            "layer": 1,
-            "weight_scale": 0.019679610218320574,
-            "membrane_scale": 0.3333333333333333,
-            "shared_scale_gauge_factor": 16.93800485047307
+            "T": 16,
+            "arm": "thresh",
+            "acc": 0.8985714285714286,
+            "syn_per_inference": 2125678.341720779,
+            "dense_per_inference": 19628032.0,
+            "event_fraction_of_dense": 0.10829808825055813
           },
           {
-            "layer": 2,
-            "weight_scale": 0.009625924485070365,
-            "membrane_scale": 0.3333333333333333,
-            "shared_scale_gauge_factor": 34.62870853083538
+            "T": 16,
+            "arm": "ftbc",
+            "acc": 0.645,
+            "syn_per_inference": 2743122.541396104,
+            "dense_per_inference": 19628032.0,
+            "event_fraction_of_dense": 0.13975535302755282
           },
           {
-            "layer": 3,
-            "weight_scale": 0.015539673822266715,
-            "membrane_scale": 0.3333333333333333,
-            "shared_scale_gauge_factor": 21.45047168594374
+            "T": 16,
+            "arm": "both",
+            "acc": 0.7207142857142858,
+            "syn_per_inference": 2544346.234577922,
+            "dense_per_inference": 19628032.0,
+            "event_fraction_of_dense": 0.12962818863235612
+          },
+          {
+            "T": 25,
+            "arm": "raw",
+            "acc": 0.8764285714285714,
+            "syn_per_inference": 2563596.2670454546,
+            "dense_per_inference": 30668800.0,
+            "event_fraction_of_dense": 0.08358971551040323
+          },
+          {
+            "T": 25,
+            "arm": "thresh",
+            "acc": 0.9285714285714286,
+            "syn_per_inference": 2243644.229707792,
+            "dense_per_inference": 30668800.0,
+            "event_fraction_of_dense": 0.07315722264020086
+          },
+          {
+            "T": 25,
+            "arm": "ftbc",
+            "acc": 0.8257142857142857,
+            "syn_per_inference": 3023309.375,
+            "dense_per_inference": 30668800.0,
+            "event_fraction_of_dense": 0.09857931758008139
+          },
+          {
+            "T": 25,
+            "arm": "both",
+            "acc": 0.81,
+            "syn_per_inference": 2707661.5560064935,
+            "dense_per_inference": 30668800.0,
+            "event_fraction_of_dense": 0.08828716989274095
+          },
+          {
+            "T": 50,
+            "arm": "raw",
+            "acc": 0.935,
+            "syn_per_inference": 2919362.8060064935,
+            "dense_per_inference": 61337600.0,
+            "event_fraction_of_dense": 0.047594995663451024
+          },
+          {
+            "T": 50,
+            "arm": "thresh",
+            "acc": 0.9371428571428572,
+            "syn_per_inference": 2426220.323051948,
+            "dense_per_inference": 61337600.0,
+            "event_fraction_of_dense": 0.039555188384481105
+          },
+          {
+            "T": 50,
+            "arm": "ftbc",
+            "acc": 0.9107142857142857,
+            "syn_per_inference": 3453336.166396104,
+            "dense_per_inference": 61337600.0,
+            "event_fraction_of_dense": 0.05630047746237388
+          },
+          {
+            "T": 50,
+            "arm": "both",
+            "acc": 0.905,
+            "syn_per_inference": 2956368.487824675,
+            "dense_per_inference": 61337600.0,
+            "event_fraction_of_dense": 0.04819830720185784
+          },
+          {
+            "T": 100,
+            "arm": "raw",
+            "acc": 0.9442857142857143,
+            "syn_per_inference": 3270968.9375,
+            "dense_per_inference": 122675200.0,
+            "event_fraction_of_dense": 0.026663652779860967
+          },
+          {
+            "T": 100,
+            "arm": "thresh",
+            "acc": 0.9292857142857143,
+            "syn_per_inference": 2701710.5844155843,
+            "dense_per_inference": 122675200.0,
+            "event_fraction_of_dense": 0.02202328249243192
+          },
+          {
+            "T": 100,
+            "arm": "ftbc",
+            "acc": 0.9342857142857143,
+            "syn_per_inference": 3888789.481331169,
+            "dense_per_inference": 122675200.0,
+            "event_fraction_of_dense": 0.031699882953776876
+          },
+          {
+            "T": 100,
+            "arm": "both",
+            "acc": 0.9207142857142857,
+            "syn_per_inference": 3266400.9139610385,
+            "dense_per_inference": 122675200.0,
+            "event_fraction_of_dense": 0.02662641604791383
           }
         ],
-        "shared_scale_note": "weight grid and membrane grid are the same grid up to a per-layer positive scalar that BatchNorm absorbs exactly; exact only with --batchnorm on",
-        "multiplier_less_compare": true,
-        "wall_clock_caveat": "this reference kernel computes in fp32 registers, so the traffic win does not appear in train_ms_per_step; a fixed-point core is where it becomes energy"
+        "calibration_reference": "same network at full T (NOT a source ANN)"
       },
       "config": {
         "method": "harmonic_matched",
@@ -1018,7 +1081,7 @@
         "decision_margin": 2.0,
         "export": false,
         "seed": 1234,
-        "hub_dir": "/kaggle/working/hub/opt08_mint_quant/u_bits_3",
+        "hub_dir": "/kaggle/working/hub/opt03_ftbc_threshold_balance/default",
         "repo_id": "UWU-R-13/SSVEP-SNN",
         "push_hf": false,
         "keep_epoch_weights": 1,
@@ -1029,149 +1092,63 @@
         "bench_warmup": 10,
         "bench_iters": 50,
         "bench_train_steps": 20,
-        "w_bits": 4,
-        "u_bits": 3,
-        "leak_shift": 0,
-        "qat": true,
-        "bits_sweep": "2,4,8,0",
-        "l2_bytes": 2097152,
+        "calibrate_thresh": true,
+        "calib_percentile": 99.9,
+        "calib_batches": 8,
+        "ftbc": true,
+        "ftbc_target_ticks": 8,
+        "ftbc_ref_ticks": 0,
+        "ftbc_lr": 1.0,
+        "t_sweep": "4,8,16,25,50,100",
         "sampling_rate": 250.0
       }
     },
     "zip": {
-      "name": "u_bits_3.zip",
+      "name": "default.zip",
       "files": 81,
-      "bytes": 12230314,
-      "MB": 12.2,
-      "sha256": "59408685a190f6abc9f857e49ce34938251b9d14b3cc1bd8192fc8a52b9a9882"
+      "bytes": 12251189,
+      "MB": 12.3,
+      "sha256": "a2408411059da5b902543faccd650a7826414bf770223e315a19896579a36f61"
     }
   },
   {
-    "key": "opt08_mint_quant/u_bits_4",
-    "tag": "u_bits_4",
-    "script": "opt08_mint_quant.py",
-    "hf_folder": "applied/opt08_mint_quant",
-    "phase": 5,
+    "key": "opt04_cudagraph_unroll/control_baseline",
+    "tag": "control_baseline",
+    "script": "opt04_cudagraph_unroll.py",
+    "hf_folder": "applied/opt04_cudagraph_unroll",
+    "phase": 1,
     "status": "ok",
-    "duration_s": 344.8,
+    "duration_s": 320.6,
     "optimization_summary": {
-      "optimization": "mint_quant",
-      "research_section": "4. Membrane-Potential Memory-Wall Optimizations (multiplier-less integer quantization)",
-      "reference": "Yin et al., MINT: Multiplier-less INTeger Quantization for Energy Efficient Spiking Neural Networks, arXiv:2305.09850 (ASP-DAC 2024)",
-      "script": "opt08_mint_quant.py",
-      "baseline_mode": false,
+      "optimization": "cudagraph_static_unroll",
+      "research_section": "2. Temporal-dimension reduction / 5. OS & runtime",
+      "reference": "CUDA Graphs static capture/replay applied to the unrolled T-tick event-driven inference loop",
+      "script": "opt04_cudagraph_unroll.py",
+      "baseline_mode": true,
       "method": "harmonic_matched",
       "accuracy": {
-        "test_acc": 0.965,
-        "test_target_acc": 0.965,
-        "best_val_acc": 0.9628571428571429
+        "test_acc": 0.9385714285714286,
+        "test_target_acc": 0.9385714285714286,
+        "best_val_acc": 0.9464285714285714
       },
       "efficiency": {
         "spikes_per_inference": null,
-        "syn_ops_per_inference": 1883576.624797078,
+        "syn_ops_per_inference": 3254834.5267857146,
         "dense_macs_per_inference": 122675200.0,
-        "event_fraction_of_dense": 0.015354176107290455,
-        "syn_ops_per_inference_dense_fanout": 64875275.42613637,
-        "weight_sparsity": 0.854853303683222,
-        "effective_ticks": 12.19,
+        "event_fraction_of_dense": 0.02653213140704653,
+        "syn_ops_per_inference_dense_fanout": 64831452.31331169,
+        "weight_sparsity": 0.7324707846410685,
+        "effective_ticks": 12.097857142857142,
         "num_ticks": 100,
-        "membrane_bytes_per_inference": 0,
-        "weight_bytes": 613376,
+        "membrane_bytes_per_inference": 451200,
+        "weight_bytes": 4907008,
         "input_raster_bytes": null,
-        "train_ms_per_step": 99.37113355001657,
-        "infer_ms_per_sample": 0.6619073024999977,
-        "peak_vram_mb": 2841.84576
+        "train_ms_per_step": 92.22967624664307,
+        "infer_ms_per_sample": 0.6272434443235397,
+        "peak_vram_mb": 2804.347904
       },
       "optimization_specific": {
-        "w_bits": 4,
-        "u_bits": 4,
-        "leak_shift": 0,
-        "beta_eff": 1.0,
-        "qat": true,
-        "batchnorm": true,
-        "membrane_memory_reduction_x": 8.0,
-        "weight_memory_reduction_x": 8.0,
-        "membrane_bytes_per_inference": 56400.0,
-        "working_set_bytes": 613940.0,
-        "baseline_working_set_bytes": 4911520.0,
-        "l2_cache_fits": true,
-        "baseline_l2_cache_fits": false,
-        "l2_budget_bytes": 2097152,
-        "membrane_saturation_frac": 0.2496060505319149,
-        "acc_vs_ubits": {
-          "2": 0.7471428571428571,
-          "4": 0.965,
-          "8": 0.9657142857142857,
-          "0": 0.9592857142857143
-        },
-        "bits_sweep_rows": [
-          {
-            "u_bits": 2,
-            "acc": 0.7471428571428571,
-            "target_acc": 0.7471428571428571,
-            "membrane_bytes_per_neuron_per_tick": 0.25,
-            "membrane_memory_reduction_x": 16.0,
-            "membrane_saturation_frac": 0.032914381094858156,
-            "acc_drop_vs_fp32": 0.2121428571428572
-          },
-          {
-            "u_bits": 4,
-            "acc": 0.965,
-            "target_acc": 0.965,
-            "membrane_bytes_per_neuron_per_tick": 0.5,
-            "membrane_memory_reduction_x": 8.0,
-            "membrane_saturation_frac": 0.2418986868351064,
-            "acc_drop_vs_fp32": -0.005714285714285672
-          },
-          {
-            "u_bits": 8,
-            "acc": 0.9657142857142857,
-            "target_acc": 0.9657142857142857,
-            "membrane_bytes_per_neuron_per_tick": 1.0,
-            "membrane_memory_reduction_x": 4.0,
-            "membrane_saturation_frac": 0.296513671875,
-            "acc_drop_vs_fp32": -0.00642857142857145
-          },
-          {
-            "u_bits": 0,
-            "acc": 0.9592857142857143,
-            "target_acc": 0.9592857142857143,
-            "membrane_bytes_per_neuron_per_tick": 4.0,
-            "membrane_memory_reduction_x": 1.0,
-            "membrane_saturation_frac": null,
-            "acc_drop_vs_fp32": 0.0
-          }
-        ],
-        "fp32_membrane_acc": 0.9592857142857143,
-        "per_layer_scales": [
-          {
-            "layer": 0,
-            "weight_scale": 0.03545519496713366,
-            "membrane_scale": 0.14285714285714285,
-            "shared_scale_gauge_factor": 4.0292302154753035
-          },
-          {
-            "layer": 1,
-            "weight_scale": 0.020786042724336897,
-            "membrane_scale": 0.14285714285714285,
-            "shared_scale_gauge_factor": 6.872743636280589
-          },
-          {
-            "layer": 2,
-            "weight_scale": 0.010473860161645072,
-            "membrane_scale": 0.14285714285714285,
-            "shared_scale_gauge_factor": 13.639397571898177
-          },
-          {
-            "layer": 3,
-            "weight_scale": 0.017392768391541073,
-            "membrane_scale": 0.14285714285714285,
-            "shared_scale_gauge_factor": 8.213594273274003
-          }
-        ],
-        "shared_scale_note": "weight grid and membrane grid are the same grid up to a per-layer positive scalar that BatchNorm absorbs exactly; exact only with --batchnorm on",
-        "multiplier_less_compare": true,
-        "wall_clock_caveat": "this reference kernel computes in fp32 registers, so the traffic win does not appear in train_ms_per_step; a fixed-point core is where it becomes energy"
+        "graph_enabled": false
       },
       "config": {
         "method": "harmonic_matched",
@@ -1211,160 +1188,124 @@
         "decision_margin": 2.0,
         "export": false,
         "seed": 1234,
-        "hub_dir": "/kaggle/working/hub/opt08_mint_quant/u_bits_4",
+        "hub_dir": "/kaggle/working/hub/opt04_cudagraph_unroll/control_baseline",
         "repo_id": "UWU-R-13/SSVEP-SNN",
         "push_hf": false,
         "keep_epoch_weights": 1,
         "no_resume": true,
         "selftest": false,
-        "baseline": false,
+        "baseline": true,
         "bench": true,
         "bench_warmup": 10,
         "bench_iters": 50,
         "bench_train_steps": 20,
-        "w_bits": 4,
-        "u_bits": 4,
-        "leak_shift": 0,
-        "qat": true,
-        "bits_sweep": "2,4,8,0",
-        "l2_bytes": 2097152,
+        "event_capacity": 0,
+        "graph_warmup": 3,
+        "bench_batches": "1,8,32,128",
         "sampling_rate": 250.0
       }
     },
     "zip": {
-      "name": "u_bits_4.zip",
-      "files": 81,
-      "bytes": 12211406,
-      "MB": 12.2,
-      "sha256": "fa14fb4378b6e3737d3d49ed8bf58b569e33a530256e925d6c0ec6d07495a44d"
+      "name": "control_baseline.zip",
+      "files": 80,
+      "bytes": 12254389,
+      "MB": 12.3,
+      "sha256": "90228451e4506fa57fd18dda54312748f259653c71dda8725406593029e0c194"
     }
   },
   {
-    "key": "opt08_mint_quant/u_bits_8",
-    "tag": "u_bits_8",
-    "script": "opt08_mint_quant.py",
-    "hf_folder": "applied/opt08_mint_quant",
-    "phase": 5,
+    "key": "opt04_cudagraph_unroll/default",
+    "tag": "default",
+    "script": "opt04_cudagraph_unroll.py",
+    "hf_folder": "applied/opt04_cudagraph_unroll",
+    "phase": 1,
     "status": "ok",
-    "duration_s": 346.7,
+    "duration_s": 501.8,
     "optimization_summary": {
-      "optimization": "mint_quant",
-      "research_section": "4. Membrane-Potential Memory-Wall Optimizations (multiplier-less integer quantization)",
-      "reference": "Yin et al., MINT: Multiplier-less INTeger Quantization for Energy Efficient Spiking Neural Networks, arXiv:2305.09850 (ASP-DAC 2024)",
-      "script": "opt08_mint_quant.py",
+      "optimization": "cudagraph_static_unroll",
+      "research_section": "2. Temporal-dimension reduction / 5. OS & runtime",
+      "reference": "CUDA Graphs static capture/replay applied to the unrolled T-tick event-driven inference loop",
+      "script": "opt04_cudagraph_unroll.py",
       "baseline_mode": false,
       "method": "harmonic_matched",
       "accuracy": {
-        "test_acc": 0.9664285714285714,
-        "test_target_acc": 0.9664285714285714,
-        "best_val_acc": 0.9628571428571429
+        "test_acc": 0.9435714285714286,
+        "test_target_acc": 0.9435714285714286,
+        "best_val_acc": 0.95
       },
       "efficiency": {
         "spikes_per_inference": null,
-        "syn_ops_per_inference": 2087678.7174310065,
+        "syn_ops_per_inference": 3255047.327922078,
         "dense_macs_per_inference": 122675200.0,
-        "event_fraction_of_dense": 0.017017936122631194,
-        "syn_ops_per_inference_dense_fanout": 65242462.345779225,
-        "weight_sparsity": 0.8476187526085142,
-        "effective_ticks": 12.227857142857143,
+        "event_fraction_of_dense": 0.026533866078246277,
+        "syn_ops_per_inference_dense_fanout": 64831665.11444805,
+        "weight_sparsity": 0.7324707846410685,
+        "effective_ticks": 11.80642857142857,
         "num_ticks": 100,
-        "membrane_bytes_per_inference": 112800,
-        "weight_bytes": 613376,
+        "membrane_bytes_per_inference": 451200,
+        "weight_bytes": 4907008,
         "input_raster_bytes": null,
-        "train_ms_per_step": 100.49617395000041,
-        "infer_ms_per_sample": 0.6706169215624413,
-        "peak_vram_mb": 2844.865024
+        "train_ms_per_step": 92.52521991729736,
+        "infer_ms_per_sample": 0.6286457926034927,
+        "peak_vram_mb": 2803.899904
       },
       "optimization_specific": {
-        "w_bits": 4,
-        "u_bits": 8,
-        "leak_shift": 0,
-        "beta_eff": 1.0,
-        "qat": true,
-        "batchnorm": true,
-        "membrane_memory_reduction_x": 4.0,
-        "weight_memory_reduction_x": 8.0,
-        "membrane_bytes_per_inference": 112800.0,
-        "working_set_bytes": 614504.0,
-        "baseline_working_set_bytes": 4911520.0,
-        "l2_cache_fits": true,
-        "baseline_l2_cache_fits": false,
-        "l2_budget_bytes": 2097152,
-        "membrane_saturation_frac": 0.32682918467420213,
-        "acc_vs_ubits": {
-          "2": 0.8592857142857143,
-          "4": 0.965,
-          "8": 0.9664285714285714,
-          "0": 0.9521428571428572
+        "graph_enabled": true,
+        "event_capacity_scan": {
+          "cap_scan": 98901,
+          "batch_size_scan": 64,
+          "n_batches_scanned": 20,
+          "density_per_sample": 1545.328125,
+          "per_layer_caps": [
+            98901,
+            7501,
+            3379,
+            3291
+          ]
         },
-        "bits_sweep_rows": [
-          {
-            "u_bits": 2,
-            "acc": 0.8592857142857143,
-            "target_acc": 0.8592857142857143,
-            "membrane_bytes_per_neuron_per_tick": 0.25,
-            "membrane_memory_reduction_x": 16.0,
-            "membrane_saturation_frac": 0.050917691710992904,
-            "acc_drop_vs_fp32": 0.09285714285714286
+        "capacity": 98901,
+        "overflow_ticks_frac": 0.0,
+        "graph_vs_eager_pred_agreement": 1.0,
+        "graph_capture_ms": 233.42132568359375,
+        "graphed_acc": 0.9435714285714286,
+        "kernel_launches_eager": 400,
+        "kernel_launches_graphed": 2,
+        "launch_reduction_x": 200.0,
+        "bench_batches": {
+          "1": {
+            "ms_per_sample_eager": 157.8002643585205,
+            "ms_per_sample_graphed": 32.571964263916016,
+            "throughput_eager_samples_per_s": 6.337124998270032,
+            "throughput_graphed_samples_per_s": 30.701249451750854,
+            "speedup_x": 4.844665279623167,
+            "capacity": 1546
           },
-          {
-            "u_bits": 4,
-            "acc": 0.965,
-            "target_acc": 0.965,
-            "membrane_bytes_per_neuron_per_tick": 0.5,
-            "membrane_memory_reduction_x": 8.0,
-            "membrane_saturation_frac": 0.26607622728280145,
-            "acc_drop_vs_fp32": -0.01285714285714279
+          "8": {
+            "ms_per_sample_eager": 20.341086983680725,
+            "ms_per_sample_graphed": 12.358421683311462,
+            "throughput_eager_samples_per_s": 49.161581227310094,
+            "throughput_graphed_samples_per_s": 80.91648153990228,
+            "speedup_x": 1.645929189416548,
+            "capacity": 12363
           },
-          {
-            "u_bits": 8,
-            "acc": 0.9664285714285714,
-            "target_acc": 0.9664285714285714,
-            "membrane_bytes_per_neuron_per_tick": 1.0,
-            "membrane_memory_reduction_x": 4.0,
-            "membrane_saturation_frac": 0.3181812389184397,
-            "acc_drop_vs_fp32": -0.014285714285714235
+          "32": {
+            "ms_per_sample_eager": 5.914023816585541,
+            "ms_per_sample_graphed": 10.751050561666489,
+            "throughput_eager_samples_per_s": 169.08961326729144,
+            "throughput_graphed_samples_per_s": 93.01416584957377,
+            "speedup_x": 0.5500879921142168,
+            "capacity": 49451
           },
-          {
-            "u_bits": 0,
-            "acc": 0.9521428571428572,
-            "target_acc": 0.9521428571428572,
-            "membrane_bytes_per_neuron_per_tick": 4.0,
-            "membrane_memory_reduction_x": 1.0,
-            "membrane_saturation_frac": null,
-            "acc_drop_vs_fp32": 0.0
+          "128": {
+            "ms_per_sample_eager": 1.9992999359965324,
+            "ms_per_sample_graphed": 9.524058550596237,
+            "throughput_eager_samples_per_s": 500.17507728351893,
+            "throughput_graphed_samples_per_s": 104.99725455146395,
+            "speedup_x": 0.20992100430455352,
+            "capacity": 197802
           }
-        ],
-        "fp32_membrane_acc": 0.9521428571428572,
-        "per_layer_scales": [
-          {
-            "layer": 0,
-            "weight_scale": 0.04109285133225577,
-            "membrane_scale": 0.007874015748031496,
-            "shared_scale_gauge_factor": 0.19161521998963355
-          },
-          {
-            "layer": 1,
-            "weight_scale": 0.020464909928185598,
-            "membrane_scale": 0.007874015748031496,
-            "shared_scale_gauge_factor": 0.38475692175839443
-          },
-          {
-            "layer": 2,
-            "weight_scale": 0.01011819179568972,
-            "membrane_scale": 0.007874015748031496,
-            "shared_scale_gauge_factor": 0.7782038438316392
-          },
-          {
-            "layer": 3,
-            "weight_scale": 0.01918845304421016,
-            "membrane_scale": 0.007874015748031496,
-            "shared_scale_gauge_factor": 0.41035177405337353
-          }
-        ],
-        "shared_scale_note": "weight grid and membrane grid are the same grid up to a per-layer positive scalar that BatchNorm absorbs exactly; exact only with --batchnorm on",
-        "multiplier_less_compare": true,
-        "wall_clock_caveat": "this reference kernel computes in fp32 registers, so the traffic win does not appear in train_ms_per_step; a fixed-point core is where it becomes energy"
+        },
+        "speedup_x_at_batch1": 4.844665279623167
       },
       "config": {
         "method": "harmonic_matched",
@@ -1404,7 +1345,7 @@
         "decision_margin": 2.0,
         "export": false,
         "seed": 1234,
-        "hub_dir": "/kaggle/working/hub/opt08_mint_quant/u_bits_8",
+        "hub_dir": "/kaggle/working/hub/opt04_cudagraph_unroll/default",
         "repo_id": "UWU-R-13/SSVEP-SNN",
         "push_hf": false,
         "keep_epoch_weights": 1,
@@ -1415,265 +1356,127 @@
         "bench_warmup": 10,
         "bench_iters": 50,
         "bench_train_steps": 20,
-        "w_bits": 4,
-        "u_bits": 8,
-        "leak_shift": 0,
-        "qat": true,
-        "bits_sweep": "2,4,8,0",
-        "l2_bytes": 2097152,
+        "event_capacity": 0,
+        "graph_warmup": 3,
+        "bench_batches": "1,8,32,128",
         "sampling_rate": 250.0
       }
     },
     "zip": {
-      "name": "u_bits_8.zip",
+      "name": "default.zip",
       "files": 81,
-      "bytes": 12213011,
-      "MB": 12.2,
-      "sha256": "63dd1fe2bcaf39fb5ab00bb833afca7ab94b1ab1e3556da5782d622e248c8bd7"
+      "bytes": 12253126,
+      "MB": 12.3,
+      "sha256": "dd13cd46e7240e145c6dd659b20ff6be55136f94e8e39058663d3ccbf1811fac"
     }
   },
   {
-    "key": "opt09_srif_softreset/lam_init_0.1",
-    "tag": "lam_init_0.1",
-    "script": "opt09_srif_softreset.py",
-    "hf_folder": "applied/opt09_srif_softreset",
-    "phase": 5,
+    "key": "opt05_bitpack_popcount/control_baseline",
+    "tag": "control_baseline",
+    "script": "opt05_bitpack_popcount.py",
+    "hf_folder": "applied/opt05_bitpack_popcount",
+    "phase": 1,
     "status": "ok",
-    "duration_s": 363.2,
+    "duration_s": 584.8,
     "optimization_summary": {
-      "optimization": "srif_softreset",
-      "research_section": "4. Membrane-Potential Memory Wall -- Soft Reset Integrate-and-Fire (SRIF), dynamic reset potential",
-      "reference": "Guo et al., 'Reducing Information Loss for Spiking Neural Networks' (SRIF), ECCV 2022; research.md S4 (stacks with MINT / SpQuant-SNN membrane quantization)",
-      "script": "opt09_srif_softreset.py",
-      "baseline_mode": false,
+      "optimization": "bitpack_popcount",
+      "research_section": "3. Spike-Sparsity & MAC->AC Restructuring",
+      "reference": "SATA, arXiv:2204.05422; BMXNet/BiFSMNv2 popcount GEMM",
+      "script": "opt05_bitpack_popcount.py",
+      "baseline_mode": true,
       "method": "harmonic_matched",
       "accuracy": {
-        "test_acc": 0.8864285714285715,
-        "test_target_acc": 0.8864285714285715,
-        "best_val_acc": 0.8864285714285715
+        "test_acc": 0.9392857142857143,
+        "test_target_acc": 0.9392857142857143,
+        "best_val_acc": 0.945
       },
       "efficiency": {
-        "spikes_per_inference": 14075.5703125,
-        "syn_ops_per_inference": 5061701.517045454,
+        "spikes_per_inference": null,
+        "syn_ops_per_inference": 3259960.8547077924,
         "dense_macs_per_inference": 122675200.0,
-        "event_fraction_of_dense": 0.04126100073238482,
-        "syn_ops_per_inference_dense_fanout": 66638319.30357143,
+        "event_fraction_of_dense": 0.02657391921682453,
+        "syn_ops_per_inference_dense_fanout": 64836578.64123377,
         "weight_sparsity": 0.7324707846410685,
-        "effective_ticks": 21.771428571428572,
+        "effective_ticks": 11.83,
         "num_ticks": 100,
         "membrane_bytes_per_inference": 451200,
         "weight_bytes": 4907008,
-        "input_raster_bytes": null,
-        "train_ms_per_step": 104.55192480001188,
-        "infer_ms_per_sample": 0.7019145781249847,
-        "peak_vram_mb": 2834.965504,
-        "discarded_charge_per_reset": -0.5145500630598341
+        "input_raster_bytes": 403200000,
+        "train_ms_per_step": 94.69575869998152,
+        "infer_ms_per_sample": 0.6265434059375252,
+        "peak_vram_mb": 2827.963904
       },
       "optimization_specific": {
-        "reset_mode": "srif",
-        "learn_reset": true,
-        "lam_init": 0.1,
-        "v_reset_init": 0.0,
-        "u_bits": 0,
-        "u_scale": 1.0,
-        "u_headroom_requested": 2.0,
-        "u_headroom_actual": 0.0,
-        "u_threshold_level": 0,
-        "u_levels_above_threshold": null,
-        "reset_family": {
-          "reset_mode": "srif",
-          "learn_reset": true,
-          "num_spiking_sites": 4,
-          "extra_params": 8,
-          "extra_params_frac_of_weights": 6.5212854757929885e-06,
-          "extra_flops_per_tick_per_neuron": 0,
-          "learned_lam_per_layer": [
-            0.34666571021080017,
-            0.329329252243042,
-            0.33023813366889954,
-            0.4990707039833069
-          ],
-          "learned_v_reset_per_layer": [
-            0.5600309371948242,
-            0.6936066150665283,
-            0.6635518074035645,
-            1.1929576396942139
-          ]
-        },
-        "reset_stats": {
-          "reset_mode": "srif",
-          "learn_reset": true,
-          "discarded_charge_per_reset": -0.5145500630598341,
-          "spikes_per_inference": 14075.5703125,
-          "syn_per_inference": 5154236.75,
-          "lam_per_site": [
-            0.34666571021080017,
-            0.329329252243042,
-            0.33023813366889954,
-            0.4990707039833069
-          ],
-          "v_reset_per_site": [
-            0.5600309371948242,
-            0.6936066150665283,
-            0.6635518074035645,
-            1.1929576396942139
-          ],
-          "u_bits": 0,
-          "membrane_saturation_frac": null,
-          "batches": 4,
-          "grid_headroom": 0.0,
-          "grid_levels_above_thresh": null
-        },
-        "reset_x_bits_sweep": {
-          "rows": [
-            {
-              "reset_mode": "hard",
-              "u_bits": 0,
-              "acc": 0.6842857142857143,
-              "target_acc": 0.6842857142857143,
-              "grid_headroom": 0.0,
-              "grid_levels_above_thresh": null,
-              "discarded_charge_per_reset": 0.15835896317720594,
-              "spikes_per_inference": 6790.640625,
-              "membrane_saturation_frac": null,
-              "membrane_memory_reduction_x": 1.0,
-              "acc_drop_vs_fp32": 0.0
-            },
-            {
-              "reset_mode": "hard",
-              "u_bits": 4,
-              "acc": 0.6864285714285714,
-              "target_acc": 0.6864285714285714,
-              "grid_headroom": 1.75,
-              "grid_levels_above_thresh": 3,
-              "discarded_charge_per_reset": 0.059665275456522254,
-              "spikes_per_inference": 7909.4375,
-              "membrane_saturation_frac": 0.22161569148936172,
-              "membrane_memory_reduction_x": 8.0,
-              "acc_drop_vs_fp32": -0.002142857142857113
-            },
-            {
-              "reset_mode": "hard",
-              "u_bits": 2,
-              "acc": 0.16214285714285714,
-              "target_acc": 0.16214285714285714,
-              "grid_headroom": 1.0,
-              "grid_levels_above_thresh": 0,
-              "discarded_charge_per_reset": 0.0,
-              "spikes_per_inference": 2683.109375,
-              "membrane_saturation_frac": 0.19872880651595745,
-              "membrane_memory_reduction_x": 16.0,
-              "acc_drop_vs_fp32": 0.5221428571428571
-            },
-            {
-              "reset_mode": "soft",
-              "u_bits": 0,
-              "acc": 0.7814285714285715,
-              "target_acc": 0.7814285714285715,
-              "grid_headroom": 0.0,
-              "grid_levels_above_thresh": null,
-              "discarded_charge_per_reset": 0.0,
-              "spikes_per_inference": 8097.6796875,
-              "membrane_saturation_frac": null,
-              "membrane_memory_reduction_x": 1.0,
-              "acc_drop_vs_fp32": 0.0
-            },
-            {
-              "reset_mode": "soft",
-              "u_bits": 4,
-              "acc": 0.7571428571428571,
-              "target_acc": 0.7571428571428571,
-              "grid_headroom": 1.75,
-              "grid_levels_above_thresh": 3,
-              "discarded_charge_per_reset": 0.0,
-              "spikes_per_inference": 8502.828125,
-              "membrane_saturation_frac": 0.2330075354609929,
-              "membrane_memory_reduction_x": 8.0,
-              "acc_drop_vs_fp32": 0.024285714285714355
-            },
-            {
-              "reset_mode": "soft",
-              "u_bits": 2,
-              "acc": 0.16214285714285714,
-              "target_acc": 0.16214285714285714,
-              "grid_headroom": 1.0,
-              "grid_levels_above_thresh": 0,
-              "discarded_charge_per_reset": 0.0,
-              "spikes_per_inference": 2683.109375,
-              "membrane_saturation_frac": 0.19872880651595745,
-              "membrane_memory_reduction_x": 16.0,
-              "acc_drop_vs_fp32": 0.6192857142857143
-            },
-            {
-              "reset_mode": "trained",
-              "u_bits": 0,
-              "acc": 0.8864285714285715,
-              "target_acc": 0.8864285714285715,
-              "grid_headroom": 0.0,
-              "grid_levels_above_thresh": null,
-              "discarded_charge_per_reset": -0.51547552848815,
-              "spikes_per_inference": 13885.140625,
-              "membrane_saturation_frac": null,
-              "membrane_memory_reduction_x": 1.0,
-              "acc_drop_vs_fp32": 0.0
-            },
-            {
-              "reset_mode": "trained",
-              "u_bits": 4,
-              "acc": 0.9092857142857143,
-              "target_acc": 0.9092857142857143,
-              "grid_headroom": 1.75,
-              "grid_levels_above_thresh": 3,
-              "discarded_charge_per_reset": -0.5824071496424404,
-              "spikes_per_inference": 17677.2734375,
-              "membrane_saturation_frac": 0.3281896193484043,
-              "membrane_memory_reduction_x": 8.0,
-              "acc_drop_vs_fp32": -0.022857142857142798
-            },
-            {
-              "reset_mode": "trained",
-              "u_bits": 2,
-              "acc": 0.06642857142857143,
-              "target_acc": 0.06642857142857143,
-              "grid_headroom": 1.0,
-              "grid_levels_above_thresh": 0,
-              "discarded_charge_per_reset": -0.6881111096608943,
-              "spikes_per_inference": 24211.9453125,
-              "membrane_saturation_frac": 0.4429418079565603,
-              "membrane_memory_reduction_x": 16.0,
-              "acc_drop_vs_fp32": 0.8200000000000001
-            }
-          ],
-          "by_mode": {
-            "hard": {
-              "fp32_acc": 0.6842857142857143,
-              "acc_by_bits": {
-                "0": 0.6842857142857143,
-                "4": 0.6864285714285714,
-                "2": 0.16214285714285714
-              }
-            },
-            "soft": {
-              "fp32_acc": 0.7814285714285715,
-              "acc_by_bits": {
-                "0": 0.7814285714285715,
-                "4": 0.7571428571428571,
-                "2": 0.16214285714285714
-              }
-            },
-            "trained": {
-              "fp32_acc": 0.8864285714285715,
-              "acc_by_bits": {
-                "0": 0.8864285714285715,
-                "4": 0.9092857142857143,
-                "2": 0.06642857142857143
-              }
-            }
+        "bitpack": false,
+        "pack_hidden": false,
+        "popcount_readout": false,
+        "n_channels": 2880,
+        "packed_words_per_tick": 90,
+        "input_raster_bytes_uint8": 403200000,
+        "input_raster_bytes_packed": null,
+        "memory_reduction_x": 1.0,
+        "mults_eliminated_per_inference": null,
+        "pred_agreement_packed_vs_float": null,
+        "popcount_counts_exact": null,
+        "crossover_density": 0.01,
+        "packed_ever_wins": true,
+        "packed_vs_scatter_ms": [
+          {
+            "density": 0.01,
+            "measured_density": 0.00997178815305233,
+            "scatter_ms": 0.1802825927734375,
+            "packed_ms": 0.046405792236328125,
+            "pack_ms": 0.1291036605834961,
+            "speedup_x": 3.8849157418824496,
+            "packed_wins": true,
+            "matches_scatter": true
+          },
+          {
+            "density": 0.05,
+            "measured_density": 0.0490993931889534,
+            "scatter_ms": 0.2356719970703125,
+            "packed_ms": 0.06339550018310547,
+            "pack_ms": 0.13286590576171875,
+            "speedup_x": 3.7174877773599095,
+            "packed_wins": true,
+            "matches_scatter": true
+          },
+          {
+            "density": 0.1,
+            "measured_density": 0.10040690004825592,
+            "scatter_ms": 0.42845726013183594,
+            "packed_ms": 0.08290290832519531,
+            "pack_ms": 0.12605667114257812,
+            "speedup_x": 5.168181295295065,
+            "packed_wins": true,
+            "matches_scatter": true
+          },
+          {
+            "density": 0.25,
+            "measured_density": 0.24960395693778992,
+            "scatter_ms": 1.0294532775878906,
+            "packed_ms": 0.14525413513183594,
+            "pack_ms": 0.12697696685791016,
+            "speedup_x": 7.087256253693126,
+            "packed_wins": true,
+            "matches_scatter": true
+          },
+          {
+            "density": 0.5,
+            "measured_density": 0.49802517890930176,
+            "scatter_ms": 1.3754749298095703,
+            "packed_ms": 0.2714681625366211,
+            "pack_ms": 0.1302051544189453,
+            "speedup_x": 5.0668001615991285,
+            "packed_wins": true,
+            "matches_scatter": true
           }
-        },
-        "mechanism": "V = u - s*((1-lam)*u + lam*thresh) + s*v_r; lam=0,v_r=0 is the baseline hard reset exactly, lam=1,v_r=0 discards zero charge",
-        "cost": "2 scalars per spiking layer; identical FMA count per tick, so the reset change is free at inference time",
-        "caveat": "the sweep swaps the reset on ALREADY-TRAINED weights, which under-states what training with each reset would give; and a charge-conserving reset can raise the spike count, so read spikes_per_inference next to the accuracy"
+        ],
+        "popcount_vs_loop_ms": {
+          "loop_ms": 0.013265609741210938,
+          "popcount_ms": 0.03826141357421875,
+          "speedup_x": 0.3467098703888335
+        }
       },
       "config": {
         "method": "harmonic_matched",
@@ -1713,34 +1516,47 @@
         "decision_margin": 2.0,
         "export": false,
         "seed": 1234,
-        "hub_dir": "/kaggle/working/hub/opt09_srif_softreset/lam_init_0.1",
+        "hub_dir": "/kaggle/working/hub/opt05_bitpack_popcount/control_baseline",
         "repo_id": "UWU-R-13/SSVEP-SNN",
         "push_hf": false,
         "keep_epoch_weights": 1,
         "no_resume": true,
         "selftest": false,
-        "reset_mode": "srif",
-        "learn_reset": true,
-        "lam_init": 0.1,
-        "v_reset_init": 0.0,
-        "u_bits": 0,
-        "u_headroom": 2.0,
-        "reset_sweep": "hard,soft,trained",
-        "bits_sweep": "0,4,2",
-        "baseline": false,
+        "baseline": true,
         "bench": true,
         "bench_warmup": 10,
         "bench_iters": 50,
         "bench_train_steps": 20,
+        "bitpack": true,
+        "pack_hidden": false,
+        "popcount_readout": true,
+        "sparsity_sweep": "0.01,0.05,0.1,0.25,0.5",
         "sampling_rate": 250.0
       }
     },
     "zip": {
-      "name": "lam_init_0.1.zip",
+      "name": "control_baseline.zip",
       "files": 81,
-      "bytes": 12307403,
+      "bytes": 12250994,
       "MB": 12.3,
-      "sha256": "f0bde8116a705a1c90f84adf38d925dec0ca1bf7025cb8bb4df284bd28d75a88"
+      "sha256": "ef07404573217d6615260626d16d6c3baa237388a674cc6f4fc50b4f57d95eb3"
+    }
+  },
+  {
+    "key": "opt05_bitpack_popcount/default",
+    "tag": "default",
+    "script": "opt05_bitpack_popcount.py",
+    "hf_folder": "applied/opt05_bitpack_popcount",
+    "phase": 1,
+    "status": "failed",
+    "duration_s": 280.1,
+    "optimization_summary": null,
+    "zip": {
+      "name": "FAILED_default.zip",
+      "files": 76,
+      "bytes": 10446958,
+      "MB": 10.4,
+      "sha256": "a05cd25014d037d0284290f3d579a1d933984716eb304384946909de3e297d93"
     }
   }
 ]