pub fn shadow_topk() -> usizeExpand description
Shadow top-k draft instrumentation (ATLAS_MTP_SHADOW_TOPK=k, default
0 = off, clamp k ≤ 8). Observational only — token selection untouched.
Each drafter forward_one D2H’s its logits (same ~200 µs the conf path
pays) and logs the top-k candidate ids + softmax probs per position;
the verify steps log the target argmaxes under the same gate. Joining
the two offline yields the per-depth conditional top-k coverage that
gates the tree-speculation build (Phase 0 of the tree-spec plan).
Value-parsed, not presence-checked (=0 really is off).
The SSOT parse. Both ModelLevers::shadow_topk (spark-model) and
SchedLevers::shadow_topk (spark-server) resolve through it once per run
rather than caching the answer in a OnceLock that a swap would pin.