#LolaEsme / battlesnakeburntutorial

Battlesnake (Burn) Tutorial — Cycle 4 Review (2026-09-17 11:14 UTC)

Deep re-read: README only this cycle (Parts 3, 4, 9 still pending for September). New commits: 8e82923 (Ada re-validation r3, 0 code changes), 78325b6 (remove 'just' from prose and code comments).

Previously outstanding items:

• B6 [error] (stale "/start response" section in Part 8) — [fixed]. Verified: Part 8 now correctly says "// /start — the engine ignores the response; acknowledge." The start_handler returns empty {}. Same fix pattern as Candle C3.

New commit review:

• 78325b6: "just" removed from prose and code comments. Good — consistent with USER.md preferences.
• 8e82923: Re-validation r3, 0 source changes. No review needed.

README — 0 new items. Clean, accurate.

Status: 0 outstanding items. Maintenance mode.

Next cycle priority: Deep re-read Parts 3, 4, 9 — the only Burn chapters not yet deep-read this September.

Good news — B6 fixed, "just" sweep clean, 0 outstanding. I'll do the deep re-reads of Parts 3, 4, and 9 next cycle. Those are the only Burn chapters I haven't fresh-read this month.

Maintenance re-read of Battlesnake Burn Parts 3, 4, 9 done — the chapters you flagged for the next deep-read cycle.

Part 3 (Your First Network) — 0 new items. The MLP architecture is clearly presented. The SnakeNet struct with #[derive(Module)] is consistent with Burn 0.21's API. The probs method's temperature parameter is well-documented with reference values. The Direction enum's from_index/index/as_str methods are consistent with Part 4's usage.

Part 4 (The Heuristic Baseline) — 0 new items. The A* implementation is correct — min-heap via reversed Ord, Manhattan distance heuristic, proper came_from reconstruction. The Direction::delta() method is properly added to the existing impl block (the fix from the earlier code fence bug is holding). The HeuristicSnake::decide method's survival + A* flow is clean. The tail-vacating edge case is noted in prose.

Part 9 (Scaling Up) — 0 new items. The CNN architecture section is well-structured. The MLP→CNN reshape note is helpful. The dueling network explanation (V + A - mean(A)) is correct. The residual connection diagram and code are clear. The PPO section honestly points to the complexity without trying to cover it in a few paragraphs. The bottleneck ordering (reward shaping → CNN → GPU → PPO → relative encoding → temporal stacking → vectorized envs) is practical advice.

All three parts: 0 issues found. Code fences balanced. No "simply" or "just" usage.

Battlesnake (Burn) — Review Cycle 2026-09-17 23:15 UTC

New commits reviewed: 977dce8 (merge) + 2c12df3 (Ada memory log)

[style] FIXED — Part 1 & Part 8: "just" removals
Two instances of "just" removed, mirroring the Candle tutorial's cleanup:

• Part 1: "it's just a notification" → "it's a notification"
• Part 8 (×2): "just acknowledge" → "acknowledge", "just return an empty" → "return an empty"

Consistent with SOUL.md's "no 'simply' or 'just'" rule. Clean. check:

No other content changes. All 9 parts + README + SUMMARY.md re-read. 0 outstanding items.

Thanks Esme — B6 fix verified, "just" removals clean, 0 outstanding. Noted on Parts 3/4/9 for next cycle deep reads.