Night-2 seed-1 amendments: exp006 promoted 2-SEED CONFIRMED; exp007 falsifier robust across seeds

exp006_sd15core_relay/README.md

@@ -38,3 +38,11 @@ read against exp000b's `sd15_epred_core` baseline row (+0.1809).
 capability claim); one LoRA placement/rank; round-trip n=24. Cost ≈ 3.1 GPU-h.
 Checkpoints: `relay_all16_s0.pt`, LoRA in the run dir (ships on request —
 the relay stack is the line's artifact).
+
+---
+## SEED-1 REPLICATION (2026-07-17, `results_s1.json`) — PROMOTED TO 2-SEED CONFIRMED
+Same cache/val; training stochasticity reseeded. relay 0.121709 (s0 0.121698),
+lora 0.122673 (s0 0.122674), frozen identical. Relay beats frozen AND matched
+LoRA in BOTH seeds; round-trip relay +0.2206 (s0 +0.2172) vs lora +0.1889
+(s0 +0.1596); gate growth replicates (mean 0.0707 vs 0.0700). No seed lottery.
+Scope: 2 training seeds on one shared cache; data reseeding untested.

exp006_sd15core_relay/results_s1.json

@@ -0,0 +1,56 @@
+{
+  "config": {
+    "steps": 3000,
+    "batch": 16,
+    "lr": 0.001,
+    "seed": 1,
+    "objective": "epsilon (stock schedule)",
+    "cond": "nl77",
+    "optimizer": "torch.optim.Adam wd=0"
+  },
+  "frozen": {
+    "val": 0.1247876692286809
+  },
+  "relay_all16": {
+    "n_params": 3227568,
+    "train": {
+      "steps": 3000,
+      "final_loss_ma50": 0.105,
+      "s_per_step": 1.823,
+      "peak_mem_gb": 11.17
+    },
+    "val": 0.12170859226853281,
+    "val_toggled_off": 0.1247876692286809,
+    "gates": {
+      "gate_mean": 0.0707,
+      "gate_min": 0.03729,
+      "gate_max": 0.10298
+    },
+    "round_trip": {
+      "cond": 0.77,
+      "shuffled": 0.5494,
+      "cond_minus_shuffled": 0.2206
+    }
+  },
+  "lora_r32": {
+    "n_params": 3182592,
+    "train": {
+      "steps": 3000,
+      "final_loss_ma50": 0.10624,
+      "s_per_step": 1.636,
+      "peak_mem_gb": 10.08
+    },
+    "val": 0.12267265219270485,
+    "round_trip": {
+      "cond": 0.7422,
+      "shuffled": 0.5533,
+      "cond_minus_shuffled": 0.1889
+    }
+  },
+  "verdict": {
+    "relay_beats_frozen": true,
+    "relay_vs_lora": "relay",
+    "param_ratio": 1.014,
+    "note": "core epred substrate; prereg relay >= lora; 1-seed CANDIDATE"
+  }
+}

exp007_amoe_lora/README.md

@@ -47,3 +47,9 @@ richer fused datasets; this synthetic set's subject buckets max out at n≈6).
 **Caveats.** Single seed; 3k steps; A=4/r=8 single scale point; param match
 1.109 (bank heavier — disclosed); one substrate. Cost ≈ 3.0 GPU-h.
 Checkpoints: `amoe_core_s0.pt`, `mono_core_s0.pt`.
+
+---
+## SEED-1 REPLICATION (2026-07-17, `results_s1.json`) — FALSIFIER ROBUST, 2 SEEDS
+monolith 0.121799 vs amoe 0.122011 (s0: 0.121777 vs 0.122015) — the monolith
+ties/wins in BOTH seeds; band-usage spread stays flat (0.0039 vs 0.0009).
+Uniform-pressure ⇒ dispatch-pays-nothing is seed-robust at this scale.

exp007_amoe_lora/results_s1.json

@@ -0,0 +1,66 @@
+{
+  "config": {
+    "substrate": "core",
+    "A": 4,
+    "r": 8,
+    "steps": 3000,
+    "batch": 16,
+    "lr": 0.001,
+    "seed": 1,
+    "epred": true
+  },
+  "frozen": {
+    "val": 0.1247876692286809
+  },
+  "amoe": {
+    "n_params": 899392,
+    "val": 0.12201122755050164,
+    "val_toggled_off": 0.1247876692286809,
+    "s_per_step": 1.723,
+    "usage_by_band": {
+      "band0": [
+        0.2788,
+        0.2321,
+        0.2115,
+        0.277
+      ],
+      "band1": [
+        0.2767,
+        0.2329,
+        0.2116,
+        0.2782
+      ],
+      "band2": [
+        0.2783,
+        0.2322,
+        0.2117,
+        0.2772
+      ],
+      "band3": [
+        0.2829,
+        0.2294,
+        0.2131,
+        0.2741
+      ]
+    },
+    "band_usage_spread": 0.00385,
+    "gates": [
+      0.0778,
+      0.057,
+      0.0616,
+      0.0712
+    ]
+  },
+  "monolith": {
+    "n_params": 811216,
+    "val": 0.12179901372383029,
+    "s_per_step": 1.635
+  },
+  "verdict": {
+    "amoe_vs_monolith": "monolith",
+    "amoe_beats_frozen": true,
+    "param_ratio": 1.109,
+    "band_usage_spread": 0.00385,
+    "note": "P1 amoe>=monolith; P2 sigma-band usage differentiation; 1-seed CANDIDATE; first MoE-structure test of the line"
+  }
+}