---
author: "Umesh Malik"
canonical: "https://umesh-malik.com/blog/tag/interpretability"
description: "Explore articles tagged with Interpretability by Umesh Malik — AI Engineer, LLM & GenAI Developer. Learn Interpretability best practices, practical tips, and in-depth guides."
title: "Umesh Malik's Blog - Interpretability Articles | Interpretability Tutorials"
tokens: 176
generator: "scripts/generate-page-markdown.mjs"
---

[← Back to Blog](https://umesh-malik.com/blog)

# Interpretability

1 article

 [![Dashboard-style cover showing disclaimer voice rising to 53 percent with a chat template versus 36 percent without one, and the steering direction that reproduces the same shift](https://umesh-malik.com/blog/how-to-stop-llm-saying-as-a-language-model-cover.png)

LLM Engineering • Sep 28, 2026

### How to Stop an LLM Saying As a Language Model: It's the Template

How to stop an LLM saying as a language model: it's the chat template, not the weights. Disclaimers drop 17 points without one — the fix, and its limits.

10 min read

Read more →](https://umesh-malik.com/blog/how-to-stop-llm-saying-as-a-language-model)