item 23 pre-registered: d=1 ceiling controls (x600 undertrain arm, pp30 pause-bandwidth arm, +-5pt decision rule); trainer gains --tag-suffix

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

results-loop/PROTOCOL_UNIFIED.md

@@ -583,3 +583,20 @@ curriculum effect + the reach trade, not scratchpad-free parity. Next
 knobs if the line continues: longer per-rung retraining, finer
 deletion schedule (fractional rungs), or E2-N noise-hardening stacked
 on the rungs.
+
+23. **E2-L d=1 capacity-ceiling controls (pre-registered 2026-07-16
+    ~15:55, before running; Nils: "run that one control").** Item 22's
+    break-at-d=1 has two untested confounds; one arm each, single-knob
+    changes from item 22's d=1 (both warm-start rung-A e400, seed 0):
+    arm A "x600" = 600 steps instead of 200 (3x training; tests the
+    brief-retrain objection — note d=1 val had plateaued at 0.36, so
+    prediction is NO recovery); arm B "pp30" = 30 pauses per deleted
+    step instead of 10 (3x latent bandwidth, same 200 steps; tests
+    whether inert-pause capacity, not training, binds). Eval n=256:
+    arm A cells 0:0, 2:12, 2:16 (e600); arm B cells 0:0, 2:32, 2:36
+    (e200). Decision rule, stated in advance: item 22 d=1 best = 31.6;
+    a control within +-5 points confirms the ceiling on that axis;
+    >=+5 (>36.6) reopens the ladder on that knob (longer schedules or
+    wider pause-chains); if BOTH land within +-5, the d=1 ceiling is
+    confirmed structural and the ladder chapter closes as scored.
+    Job: scripts/jobs/zzz_n_rungb_ctrl.sh.

scripts/jobs/zzz_n_rungb_ctrl.sh

@@ -0,0 +1,13 @@
+# gpuq-in:  results-loop/star_data.json results-loop/gsm_cot_data.json results-loop/adapter_carrycot_e400.pt
+# gpuq-out: results-loop/eval_gsm_carrycot_b1_x600.json results-loop/eval_gsm_carrycot_b1_pp30.json results-loop/train_carrycot_b1_x600_log.json results-loop/train_carrycot_b1_pp30_log.json results-loop/adapter_carrycot_b1_x600_*.pt results-loop/adapter_carrycot_b1_pp30_*.pt
+git pull origin main -q 2>/dev/null
+P=/home/nils/jspace/.venv/bin/python
+export JLENS_MODEL=google/gemma-4-E2B-it LOOP_OUT=/home/nils/jspace/results-loop
+cd /home/nils/jspace/scripts
+WARM=$LOOP_OUT/adapter_carrycot_e400.pt
+# arm A (undertrain control): d=1, 3x the training, same 10 pauses
+$P train_carry_cot.py --drop-steps 1 --warm-start $WARM --steps 600 --lr 3e-4 --tag-suffix _x600
+$P eval_carry_cot.py --adapter $LOOP_OUT/adapter_carrycot_b1_x600_e600.pt --tag gsm_carrycot_b1_x600 --grid 0:0,2:12,2:16 --n 256
+# arm B (pause-bandwidth control): d=1, same 200 steps, 3x the pauses
+$P train_carry_cot.py --drop-steps 1 --warm-start $WARM --steps 200 --lr 3e-4 --pause-per-step 30 --tag-suffix _pp30
+$P eval_carry_cot.py --adapter $LOOP_OUT/adapter_carrycot_b1_pp30_e200.pt --tag gsm_carrycot_b1_pp30 --grid 0:0,2:32,2:36 --n 256

scripts/train_carry_cot.py

@@ -51,12 +51,14 @@ ap.add_argument("--pause-per-step", type=int, default=10,
 ap.add_argument("--warm-start", default=None, metavar="ADAPTER_PT")
 ap.add_argument("--steps", type=int, default=STEPS)
 ap.add_argument("--lr", type=float, default=LR)
+ap.add_argument("--tag-suffix", default="",
+                help="appended to TAG (distinguish control variants)")
 ARGS = ap.parse_args()
 STEPS, LR = ARGS.steps, ARGS.lr
 TAG = ("carrycot_ff" if ARGS.feedforward else "carrycot") + (
     f"_b{ARGS.drop_steps}" if ARGS.drop_steps else "") + (
     f"_ln{ARGS.lensnoise.replace(',', '_')}" if ARGS.lensnoise else "") + (
-    f"_s{ARGS.seed}" if ARGS.seed else "")
+    f"_s{ARGS.seed}" if ARGS.seed else "") + ARGS.tag_suffix
 
 
 def drop_cot_steps(cot, d):