Stories by nkko

AI Sandbox Benchmark

Superposition of Features Creates Power Law Performance in LLMs

LLM inference economics from first principles

Intellect-2: A Reasoning Model Trained Through Globally Decentralized RL

Qwen3 Technical Report [pdf]

Show HN: Hacker News Karma Tracker GitHub Action

LegoGPT: Generating Physically Stable and Buildable Lego

Model Soup's Recipe for Embeddings

Base Models Beat Aligned Models at Randomness and Creativity

Seductive Language for Narcissists in Job Postings

Recursive Planning Based Open Writing Project: WriteHERE

OSS Customer support agents by Antiwork

Real-time collaborative stream overlay

Evaluating Frontier Models for Stealth and Situational Awareness

Llama-Nemotron: Efficient Reasoning Models

Dummy's Guide to Modern LLM Sampling

Physics of Language Models: Architecture Design and the Magic of Canon Layers

Groq's First Compound AI System (LLM with Compute)

Qwen3: Supervised Fine-Tuning with TRL

1
2
3
4
5