---
name: undominated-quant-degradation-audit
description: Audits perplexity and benchmark score degradation across quantization levels (BF16, FP8, INT4 AWQ, Q4_K_M) to prevent claiming unquantized performance for quantized serving.
license: MIT
metadata:
  author: Undominated.ai
  version: "1.0.0"
  tags: ["quantization", "benchmarks", "fp8", "awq", "fidelity"]
---

# Quantization Degradation Audit

Audits perplexity and benchmark score degradation across quantization levels (BF16, FP8, INT4 AWQ, Q4_K_M) to prevent claiming unquantized performance for quantized serving.

## When to Use
Use when evaluating low-cost inference provider endpoints serving quantized models.

## Limitations
Scores must derive from identical evaluation harness configurations and matching temperature.

## Verification Script
Run the deterministic stdlib checker:
```bash
python3 scripts/check.py examples/synthetic.json
```
Exits `0` on verified pass, `1` on failed assertion, `2` on malformed input.
