tiny neural network

Checkpoint: CP-004 Status: QUEUED — not yet run. No results exist for this experiment.

Question

What does it cost to prove Linear -> ReLU -> Linear end to end, and which layer class dominates?

Why it matters

First workload with a non-linearity. ZIP's Table 7 predicts linear layers dominate; confirming or refuting that at our own scale decides whether specialised activation proofs are worth pursuing at all.

Gate

Blocked on CP-002 and CP-003.

---

This is a stub. It exists so the roadmap's structure is visible, not to imply work has been done. Nothing in ../../benchmarks/results/ refers to it yet.