harsh.
  • Blog
Blog

Thoughts & Writing

Blogs on various topics.

January 18, 2026

Understanding Recursive Language Models: Managing Long Contexts

A deep dive into RLMs - a new inference paradigm that lets LLMs process millions of tokens by treating context as a part of an external environment which a LLM can interact with. Includes a minimal implementation from scratch.

Read more

© 2026 Harsh Jawajiwar