---
title: Is the small survival network as accurate as it can be?
family: lifetime-objective
summary: Retrains the small board evaluator at several sizes and settings to test whether more capacity predicts lifetime better.
status: completed
evidence: repository-verified
reads: teacher
---

Retrain the tiny board-scoring network that already fits inside the search at
several widths, training lengths and learning rates, and ask whether any of
them predicts remaining lifetime better than the one that played — without
playing a single game.

<EvidenceLabel status="completed" evidence="repository-verified" reads="teacher" />

## What happened

Nothing beat the network that already played. Across thirteen retrainings the
best configuration was 0.0022 more accurate than the baseline, and the three
training seeds of the baseline itself spread by 0.0023, so the difference is
training noise, and the gate fixed in advance reads *supported-as-tested*: the
network's accuracy on this corpus is saturated. The revealing part is the
shape of the failure. Making the network wider lowered its training loss step
by step and lowered its held-out accuracy step by step; the eight-times wider
version was the worst predictor of games it had not seen. That is a network
memorising a corpus that is too small for it, not one that is too small for
the job. The next learned-leaf experiment should change the data or the
target, not the width.

<TechnicalDetails title="The result">

Result `RS-20260822T024228Z-94090db1` for `EX-20260822-nnue-leaf-capacity-sweep-79632fd9`, run
`RUN-20260822T020655Z-feaf4331`, valid, pass (supported-as-tested), evidence
tier mechanics-only (offline, no game played). Held-out lifetime Pearson on
the whole-origin test split (486,819 examples): baseline h64/m32/10 epochs
0.8546 (seeds 42283 and 42284: 0.8567, 0.8569; spread 0.0023); h32 0.8550;
h128 0.8543; h256 0.8515; h256/m64/20 epochs 0.8446; h512/m64/20 epochs
0.8378 (final training loss 0.5110 → 0.4197 over that ladder); h64 at 20
epochs 0.8564; mid 16 / mid 64 0.8542 / 0.8568; lr 1e-3 0.8552. Mirrored-board
Pearson matches each row within 0.0004. The recorded finding-08 baseline was
0.8564 on an earlier torch build. Full table in `runs/RUN-20260822T020655Z-feaf4331/summary.json`.

</TechnicalDetails>

<Callout title="An offline diagnostic, not a strength result" tone="warn">
This is an offline diagnostic. Its labels come from the futures of already
played training games (a teacher signal), and its metric is how well a network
predicts those futures on games it never saw. It says nothing about playing
strength: the [learned-leaf experiment](/approaches/lifetime-objective/learned-leaf)
already found that a 0.008 gap in this metric did not separate two models in
play. What it can settle is narrower, whether the network's accuracy is
limited by its size and training, or by what it is shown.
</Callout>

## The intuition

The [learned leaf](/approaches/lifetime-objective/learned-leaf) is the one
learned evaluator in this repository that helped inside the search, and its
page ends with a claim: "train a bigger survival model is not the next
experiment", because a network five times its size was only 0.008 more
accurate. That claim rests on a single comparison between two quite different
architectures. If the small network's accuracy is set by its *inputs* — 135
active features per board, mostly single cells and adjacent pairs — then
making it wider or training it longer will not help, and the next learned-leaf
experiment should change what the network sees or what it is asked to predict.
If instead the network is simply too small or under-trained, wider versions
will be more accurate, and the claim is wrong.

Each training run costs about half a minute on the integrated GPU and reads
only the corpus that already exists, so the question is cheap to settle
either way.

## How it works, step by step

1. **Same data, same split.** Every run trains on the existing
   5.26-million-position corpus with the whole-origin split the learned-leaf
   work used, so held-out numbers are comparable with the recorded baseline.
2. **A small grid.** Hidden width 32 to 512, middle layer 16 to 64, ten or
   twenty epochs, two learning rates — thirteen runs, including three training
   seeds of the baseline so that a difference can be compared with the noise
   of training itself.
3. **One number per run.** Held-out correlation between predicted and actual
   remaining lifetime, read straight from the trainer's own output file;
   mean error in moves, hazard accuracy and the mirrored-board check are
   reported beside it.
4. **A gate fixed in advance.** The claim "saturated" is supported if the best
   run beats the baseline by no more than twice the seed-to-seed spread (or
   0.005, whichever is larger); it is rejected if the best run is at least
   0.02 better and more than three spreads away; anything between is
   inconclusive.

<TechnicalDetails title="The record">

- Theory `TH-20260822-nnue-leaf-capacity-saturation-9c45e42e`; experiment
  `EX-20260822-nnue-leaf-capacity-sweep-79632fd9` (frozen, CHECK tier,
  diagnostic, no seed opened).
- Trainer: `approaches/lifetime-objective/learned-leaf/train_leaf.py`,
  unchanged; corpus `runs/RUN-A51D-corpus/all.states`; the recorded baseline
  is held-out lifetime Pearson 0.8564 (`runs/RUN-A52-LEAF/model/leafnet-h64.json`).
- `sweep.sh` runs the grid; `summarize.py` tabulates and applies the gate.
- Limitation stated in advance: inference cost per board is not measured here,
  so a wider network that wins offline is not yet known to fit the roughly
  one-microsecond leaf budget the search imposes.

</TechnicalDetails>

## Sources

- `sweep.sh`: the thirteen-run grid, reading the existing corpus only
- `summarize.py`: the table and the preregistered gate