Notes on Interpreting LLMs

Introduction

This is a placeholder post that demonstrates the blog layout and features.

• Written in MDX
• Supports code blocks, links, and lists
• Full-text search and tag filtering

Code Example

def attention(q, k, v):
    scores = q @ k.T / (k.shape[-1] ** 0.5)
    return softmax(scores) @ v

Quote Example

Interpretability and safety are critical directions in modern AI research.

Replace this post with your real content once ready.