---
id: 7016
title: "Lean and Mean: How We Fine-Tuned a Small Language Model for Secret Detection in Code"
source_domain: wiz.io
source_url: "https://www.wiz.io/blog/small-language-model-for-secrets-detection-in-code"
published_at: "2025-06-10T16:01:49Z"
tags: ["ai-security", "llm", "machine-learning", "code-security", "secrets-detection", "wiz-research", "small-language-model"]
phying_url: "https://phying.com/item/7016"
lang: en
---

# Lean and Mean: How We Fine-Tuned a Small Language Model for Secret Detection in Code

## Summary

Wiz engineers fine-tuned a 1B-parameter Llama model to spot secrets in code, hitting 86% precision and 82% recall on standard CPUs — beating regex without the cost and privacy issues of large LLMs.

## Intelligence Metadata

- **Sources:** wiz.io
- **Published:** 2025-06-10T16:01:49Z
- **Original:** https://www.wiz.io/blog/small-language-model-for-secrets-detection-in-code
- **Phying News:** https://phying.com/item/7016