# Multi-Token Prediction Speedup Audit

## Description
Calculates acceptance rate thresholds and net speedup of Multi-Token Prediction (MTP) heads vs standard autoregression.

## Category
`serving`

## Tags
`mtp`, `multi-token-prediction`, `speculative`, `speedup`, `serving`

## When to Use
Audit models equipped with native Multi-Token Prediction heads (e.g. DeepSeek-V3).

## Operational Limitations
Evaluates observed acceptance rate and head compute cost.

## Verification Script
Run the deterministic verification checker:
```bash
python3 scripts/check.py examples/synthetic.json
```
Exit codes:
- `0`: Check passed and thresholds satisfied.
- `1`: Threshold breach or operational violation detected.
- `2`: Schema error or invalid usage.