item 27 pre-registered: zero-pause internal band looping — single M=10 in-place burst at prompt end, iteration-aligned lens targets, ii0 inference ablation; carry_steps gains inplace updates, trainer gains --inner-iters/--base-pauses

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

scripts/eval_carry_cot.py

@@ -38,6 +38,8 @@ def main():
     ap.add_argument("--max-new", type=int, default=160)
     ap.add_argument("--bandlora", default=None, metavar="LORA_PT",
                     help="load a lora_*_e*.pt loop-only band-LoRA checkpoint")
+    ap.add_argument("--inner-iters", type=int, default=0, metavar="M",
+                    help="M in-place band iterations at the last pause")
     args = ap.parse_args()
 
     model, tok = load_model(dtype=torch.bfloat16)
@@ -83,7 +85,8 @@ def main():
                                        enc["input_ids"],
                                        enc["attention_mask"], k, p,
                                        max_new_tokens=args.max_new,
-                                       feedforward=args.feedforward)
+                                       feedforward=args.feedforward,
+                                       inner_iters=args.inner_iters)
             for j, it in enumerate(chunk):
                 txt = tok.decode(gen[j, enc["input_ids"].shape[1]:],
                                  skip_special_tokens=True)