Table 1The policy-only MLP used in the experiments

Empirical result Supported by saved experimental runs, not by a proof.

1 Paper text

Component	Type	Dimensions	Activation/Normalisation
Input	State representation	Two-frame: $2Nk$ bits (1F: $Nk$ )	None
Hidden 1	Fully Connected	1F: $Nk\to H$ ; 2F: $2Nk\to H$	LayerNorm + ReLU
Hidden 2	Fully Connected	$H\to H$	ReLU
Policy Head	Linear Output	$H\to N\cdot2^k$	Raw logits; Log-Softmax in loss

Table 1: Neural network architecture used in the experiments.

H

denotes the hidden width (a fixed hyperparameter).

Generated from paper_jmlr.tex, lines 1012–1027.

2 Explain this

Both restoration agents use the same small multilayer perceptron: input, two hidden layers of width $H$ (the first with LayerNorm and ReLU, the second with ReLU) and a linear output of $N\cdot2^k$ logits. There is no value head. The 2F model has a wider input ( $2Nk$ bits) and therefore $NkH$ more first-layer parameters (20,480 in the reported setting).

3 How this is formalised

How this is covered. An empirical setup record from the saved notebook; not formalised. These are ordinary floating-point networks, not instances of the fixed-scale gate model.

Coverage record
Paper anchor	`table:arch` (manuscript label)
Displayed as	Table 1
Claim	The policy-only MLP used in the experiments
Coverage category	Empirical result
Statement source	paper_jmlr.tex lines 1012–1027; SHA-256 of the normalised block source `0463fce4170d5fb3…`
Explanations	`explain/table:arch`, `formal/table:arch`
Build evidence	fresh isolated rebuild, audit report

4 Lean statement and source

No Lean statement corresponds to this claim. This is an empirical statement about saved experimental runs. Lean proofs do not certify training runs, curves or checkpoints.

Build evidence: fresh isolated rebuild of all 38 bundled modules with Lean (version 4.26.0, started 2026-10-10T15:21:56Z, finished 2026-10-10T15:27:00Z: 38/38 modules compiled, 1588 axiom reports, no errors. See Build for the commands and the independent audit of 2026-10-10 (dated evidence).