Hierarchos: Preliminary Findings From a 232M Recurrent Memory-Augmented Assistant Model [P]
Researchers released preliminary findings for Hierarchos, a 232M-parameter recurrent memory-augmented language model, detailing its architecture and training results.
Project Release / Research Draft] Hierarchos at 232M Parameters: Preliminary Findings From a Recurrent Memory-Augmented Assistant Model
Technical Report: July 2nd, 2026
Project: Hierarchos / KortexHOS
Authors: Makhi Burroughs / netcat420, Lost Time, and the Hierarchos project team
TL;DR:
We built and trained Hierarchos, an experimental 232M-parameter recurrent, memory-augmented language model
Source: Hierarchos: Preliminary Findings From a 232M Recurrent Memory-Augmented Assistant Model [P]. Read the full piece at the source.
Summary and analysis generated by AI (mistral). Always verify against the original sources.
Measuring the Economic Effects of AI - Economic Innovation Group
Claude Code and China: The mechanism is activated when the user sets the ANTHROPIC_BASE_URL environment variable (used for local models)
llamacpp patch - DeepSeek V4 Flash running with full 1M token context locally on RTX 5090
