Blog
Notes, ideas, and technical write-ups.
A technical map of AI4AI and recursive self-improvement: system loops, the Gödel Machine lineage, evaluation standards, and the teams advancing bounded RSI.
An evidence-led overview of recurrent depth, from model-loop and layer-loop pretraining to stability, scaling ladders, and test-time adaptation.
Starting from an interview claim: a systematic test with open-weight reasoning models.