AI-native strategy · Claude Code · GitHub Copilot training →
MWModern Web Labs
ServicesMethodTutorialsInsightsLab
Start an Adoption ConversationKO
  • Services
  • Method
  • Tutorials
  • Insights
  • Lab

Lab

Inside LLMs, Visualized

A hands-on collection of interactive aids for building intuition about context and tokens.

Share
↓

Understand why attention dilutes (Part 1), then see the practical traps that follow (Part 2). The tokenizer is a standalone tool you can jump into anytime.

Part 1 · Cause (Mechanism)

Why attention disperses as tokens grow

Transformer attention computes token-to-token relations at N² cost. See how attention dilutes as the token count grows.

Attention · N² Scaling

Part 2 · Consequence

Lost in the Middle: position-based accuracy curve

Find out why an LLM forgets the instructions you gave at the start as a conversation grows, explained through the accuracy-by-position curve.

Position Bias · Smart Zone

Tool · Tokenizer

Tiktokenizer: how a tokenizer splits text into tokens

Type text and watch how each model's tokenizer draws token boundaries in real time. An interactive tool spanning token counts, IDs, and embeddings.

Tokenization · Interactive

Modern Web Labs

Modern Web Labs is an Official Anthropic Partner. AI-native engineering consulting and Claude Code training for high-performing teams.

ServicesMethodTutorialsInsightsLab