Continual MI · Blog
Notes from the build
Engineering and research notes on how we build MGPT, MDL, and the rest of the path toward continual machine intelligence.
MDLEngineeringGenerationCost
Block Generation: How We Made Endless Visual Novels Affordable
MDL's endless visual novels were immersive but expensive. The fix came from a reverted training experiment and a setting nobody re-enabled: one model call now writes a whole block of story beats instead of one.
Read postMGPTResearchBenchmarkPaper
Bounded-Memory Long-Horizon Conversations: MGPT on BABILong
A 4B-parameter MGPT reaches 99% token accuracy on BABILong QA1 with constant memory usage, by letting the model govern its own KV cache through masking. The demo, the result, and the paper.
Read post