Where should the adaptive reasoning loop live: tokens, harness, or weights? A recursive-LLM line (HRM, TRM, GRAM) on internalizing the AGI kernel.
Read Post
AGI Envision: from the monolithic Singleton to a System-level architecture. The open question of whether ARC-AGI-3 and starting a company require the shift.
Read Post
The environment is no longer a passive test harness. It is a data engine. 10 dimensions of scaling, from task generation to multi-agent self-play.
Read Post
13 interactive puzzle games with 1,872 levels inspired by The Witness, compatible with ARC-AGI-3 SDK and RL-ready via OpenEnv — an open-source training ground for teaching machines fluid intelligence
Read Post
Re-examining mid-training as the strategic centerpiece of the LLM pipeline — how it builds the knowledge foundation for agentic RL, and how RL signals are now flowing backward to improve mid-training itself
Read Post
A Step-by-Step Walkthrough using Slime and SWE-Bench as an Example
Read Post
From GPU Memory Budgets to Framework Architectures
Read Post
Low-Resource RLVR for Transforming Instruct Models into Reasoning Models
Read Post
Efficient Architectures for Agentic Tasks
Read Post