Archive MaleCNS research sources, provenance, bibliography and immutable data inventory

2026-09-29 03:38:32 +08:00
commit 3981628ff8
682 changed files with 161475 additions and 0 deletions
@@ -0,0 +1,25 @@
+phase: 4N
+name: MaleCNS-CreditAssignment-audit
+frozen_phase4m_config: /data/digital-drosophila/models/male-cns-lif/kcmbon-efficacy-v1/phase4m.yaml
+experiment_root: /data/digital-drosophila/experiments/phase4n-credit-assignment
+model_root: /data/digital-drosophila/models/male-cns-lif/credit-assignment-v1
+mechanism: bounded_multiplicative
+checkpoints: [0, 1, 2, 5, 10, 20, 30, 40, 50, 60, 75, 100]
+train_seeds: [51201, 51202, 51203, 51204, 51205]
+evaluation_seeds: [51301, 51302, 51303, 51304, 51305]
+edge_identity_source: phase4l matched_energy held-out ODOR_A/B majority-of-five KC representation
+credit_specificity_index: group mean anatomical-weighted E depression for A-specific minus maximum of shared and B-specific groups
+functional_separation: A MBON spike-count L1 change minus B MBON spike-count L1 change; first-spike latency used only as tie-breaker
+peak_rule: earliest checkpoint with at least 3 of 5 A-specific functional seeds and maximum five-seed mean spike-L1 separation; spike-L1 ties within 1e-9 use mean first-latency separation only if difference exceeds half a dt step (0.05 ms), otherwise choose earliest; no peak if no such checkpoint
+generalization_onset_rule: earliest checkpoint after peak with CSI at least 20 percent below peak and shared-or-B group depression at least 10 percent above its peak value; otherwise unidentified
+random_us_seed_offset: 700000
+random_us_window_ms: [30.0, 110.0]
+shadow_trace_tau_ms: 20.0
+event_coincidence_window_ms: 1.0
+first_activation_only: true
+matched_compartment_verbatim: "y5B'2a"
+unrelated_compartment_verbatim: "y2a'1"
+matched_dan_type: PAM15
+unrelated_dan_type: PPL103
+no_frozen_artifact_writes: true
+no_formal_associative_memory_claim: true