release: daydream-chess-nanogpt-micro-1 (checkpoint + ONNX + tokenizer + model card)

README.md

@@ -0,0 +1,133 @@
+---
+license: mit
+language:
+  - en
+library_name: nanogpt
+pipeline_tag: text-generation
+tags:
+  - daydream
+  - nanogpt
+  - char-level
+  - gpt
+  - chess
+  - uci
+  - minichess
+---
+
+# Model Card — `daydream-chess-nanogpt-micro-1` (v1, Micro)
+
+> A [sup computer](https://supcpu.romellogoodman.com) release — a small language model studio. [Model page](https://supcpu.romellogoodman.com/models/daydream-chess-nanogpt-micro-1/) · [monorepo](https://github.com/romellogoodman/sup-computer) (frozen code: [`projects/daydream/models/daydream-chess-nanogpt-micro-1/`](https://github.com/romellogoodman/sup-computer/tree/main/projects/daydream/models/daydream-chess-nanogpt-micro-1), tag `daydream-chess-nanogpt-micro-1`) · runs in your browser at [supcpu.romellogoodman.com/model-player](https://supcpu.romellogoodman.com/model-player/).
+
+
+<div class="takeaways">
+<p class="takeaways-label">Key takeaways</p>
+<ul>
+<li>A <strong>0.79M-param</strong> char-level GPT trained entirely on <strong>synthetic self-play</strong> — no human corpus exists for 5×5 Gardner minichess, so all 4,135 training games came from two Fairy-Stockfish instances playing each other.</li>
+<li>Fixed-depth engine self-play is <strong>fully deterministic</strong> on its own — the first generation attempt produced identical games every time. Fixed by randomizing opening plies (sourced from the engine's own legal-move list) before search takes over.</li>
+<li><strong>100% clean completion, 39.2% legal-move rate</strong> on first try — slightly higher than the <a href="daydream-chess-nanogpt-1.md">Regular</a> tier's 35.3%, consistent with a smaller board being an easier legality problem to learn, though the corpora and vocab sizes differ too much to call it a controlled comparison.</li>
+<li>Smallest tier in the three-board <a href="../../projects/daydream/README.md">daydream</a> family — 5×5 is the smallest board that can hold one of every standard chess piece, which is why Micro uses Gardner's real, balance-tested arrangement rather than an invented one.</li>
+</ul>
+</div>
+
+The smallest tier in the [`daydream`](https://github.com/romellogoodman/sup-computer/blob/main/projects/daydream/README.md)
+family: a chess-move GPT trained on **Gardner minichess**, a real 5×5 chess
+variant — one each of King/Queen/Rook/Bishop/Knight per side, five pawns.
+Same mechanic as the rest of the series: legal moves snap into focus,
+illegal moves render as dim near-misses instead of being discarded.
+
+> **A smaller board means a smaller book to memorize.** The animating
+> thesis behind the daydream series is that repetition (opening theory,
+> memorized lines) is where a model is most "in focus" and least
+> interesting. Micro tests the far end of that: with only 25 squares and 6
+> non-pawn pieces per side, there's very little room for memorized
+> structure at all — almost everything the model does here, it has to
+> generalize from a comparatively tiny, self-play-only corpus.
+
+## Model details
+
+| | |
+|---|---|
+| **Version / git tag** | `daydream-chess-nanogpt-micro-1` (research run `micro-r1`) |
+| **Architecture** | modern char-level (RoPE, RMSNorm, bias-free) on the shared `core` engine |
+| **Size** | 4 layers · 4 heads · 128 embedding dim · 128 context · dropout 0.1 · **~0.79M params** |
+| **Tokenizer** | character-level, **15-char** vocabulary over UCI move text on a 5×5 board (files a–e, ranks 1–5, promotion letters n/q/r, space, newline) |
+| **Checkpoint** | `projects/daydream/models/daydream-chess-nanogpt-micro-1/` (weights not committed) |
+| **Built on** | the monorepo's shared [`core`](https://github.com/romellogoodman/sup-computer/tree/main/core) engine |
+| **Developed with** | Claude ([Claude Code](https://claude.com/claude-code)) |
+| **License** | MIT |
+
+## Intended use
+
+Same exhibit posture as Regular, scaled to the smallest board in the
+series. Pairs with `harness.py` (this folder), which plays the model
+against Fairy-Stockfish under the built-in `gardner` variant.
+
+**Out of scope.** Not a chess engine, not evaluated for playing strength.
+Vocabulary and board are Gardner-minichess-specific — moves here are
+meaningless on Regular's or Grand's boards and vice versa (see
+[ADR-0022](https://github.com/romellogoodman/sup-computer/blob/main/docs/adr/0022-daydream-three-tier-sampler-prober-shape.md)
+on why tiers never share a vocabulary).
+
+## Training data
+
+No human corpus exists for 5×5 chess, so this tier is entirely synthetic:
+**4,135 self-play games** between two Fairy-Stockfish instances under the
+engine's built-in `gardner` variant (bounded-depth search, not
+strength-reduced — see
+[ADR-0021](https://github.com/romellogoodman/sup-computer/blob/main/docs/adr/0021-daydream-fairy-stockfish-dependency.md)),
+with randomized opening plies for game-to-game diversity (fixed-depth
+search alone is fully deterministic and produced identical games on the
+first attempt — fixed by sourcing random legal openings from the engine's
+own `go perft 1` move list) plus a repetition-window cutoff for games that
+fell into shuffling loops. Corpus is vendored in-folder as `games.txt`
+(synthetic, seeded, code-owned — committed, same treatment as
+kenosha-kid's `raw.txt`).
+
+## Training procedure
+
+- **Optimizer:** AdamW, LR 3e-4 with cosine decay to 3e-5, 100 warmup iters, β₂ 0.99, batch size 64.
+- **Run:** 2,500 iterations, best val loss **0.718**.
+- **Hardware:** Apple Silicon Mac (MPS / Metal backend), `torch.compile` disabled.
+
+## Evaluation
+
+| Metric | Result (30 games) |
+|---|---|
+| **Clean completion rate** | 30/30 (100%) |
+| **Legal-move rate (first try)** | 121/309 (39.2%) |
+
+Micro's legal-move rate (39.2%) is somewhat higher than Regular's (35.3%,
+[`daydream-chess-nanogpt-1`](https://github.com/romellogoodman/sup-computer/blob/main/research-docs/model-cards/daydream-chess-nanogpt-1.md)) — consistent
+with a smaller board and smaller per-position legal-move count being an
+easier legality-learning problem, though the two aren't a strict
+apples-to-apples comparison (different corpora, different vocab sizes,
+different training run lengths).
+
+## Limitations
+
+- **Not evaluated for playing strength**, deliberately.
+- **Synthetic corpus only** — no human Gardner-minichess games exist to
+  compare against; the training distribution is entirely a product of
+  bounded-depth Fairy-Stockfish self-play plus randomized openings.
+- **Legality is learned, not guaranteed** — same resample-then-force-random
+  fallback as every tier in this series.
+- **No weights in the tree** ([ADR-0002](https://github.com/romellogoodman/sup-computer/blob/main/docs/adr/0002-no-weights-in-tree.md)).
+
+## How to reproduce
+
+```bash
+cd projects/daydream/models/daydream-chess-nanogpt-micro-1
+python prepare.py             # -> micro/{train,val}.bin + meta.pkl
+python train.py config.py     # -> ./ckpt.pt (2500 iters, val ~0.72)
+python harness.py --games 30  # verification
+```
+
+Requires Fairy-Stockfish on `PATH` (`brew install fairy-stockfish`).
+
+Experiment write-up: [Can a chess model's illegal moves be the point?](https://github.com/romellogoodman/sup-computer/blob/main/research-docs/reports/illegal-moves-are-the-point.md)
+
+## Citation / credits
+
+- The shared `core` engine (modern nanoGPT lineage — RoPE, RMSNorm, bias-free).
+- [Fairy-Stockfish](https://github.com/fairy-stockfish/Fairy-Stockfish) — self-play corpus generator and legality arbiter, via its built-in `gardner` variant.
+- Set up and trained with Claude ([Claude Code](https://claude.com/claude-code)).

ckpt.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c1eea5e6937d378e6b1d86563e374e3314df4069ab78605982e598fa30b6860
+size 9508438

daydream-chess-nanogpt-micro-1.int8.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:007f1f07eb2682010ac272af73f20b7d64b2eeef82687155114fed47497b68ce
+size 1019493

daydream-chess-nanogpt-micro-1.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f67db1e11dedb23d80220e3dce46fffd4fd6bafdf7cef57eee87b19d333b1b1
+size 3312555

daydream-chess-nanogpt-micro-1.vocab.json

@@ -0,0 +1 @@
+{"stoi": {"\n": 0, " ": 1, "1": 2, "2": 3, "3": 4, "4": 5, "5": 6, "a": 7, "b": 8, "c": 9, "d": 10, "e": 11, "n": 12, "q": 13, "r": 14}, "itos": {"0": "\n", "1": " ", "2": "1", "3": "2", "4": "3", "5": "4", "6": "5", "7": "a", "8": "b", "9": "c", "10": "d", "11": "e", "12": "n", "13": "q", "14": "r"}}