AI & ML interests

UIUX, augment, augmentation, blockchain, enhancement,knowledge,economy

Recent Activity

Gregory-L  updated a dataset about 12 hours ago
PYTHAI/mindXascension
Gregory-L  updated a Space about 13 hours ago
PYTHAI/minaiml
Gregory-L  published a Space about 13 hours ago
PYTHAI/minaiml
View all activity

Organization Card

PYTHAI

python augmented intelligence

PYTHAI (say it Pyth-A-I) is python augmented intelligence: the intersection of blockchain and AI, built in the open. PYTHAI suggests that intelligence is, or intelligence is not. In our work, AI means augmented intelligence, and machine learning is the extraction of knowledge from information.

Our flagship is mindX, a self-building cognitive architecture. Its Belief–Desire–Intention agents audit themselves against eight Gödel-machine predicates, dream over their own memory, and train their own weights from those dreams on a two-core CPU with no GPU. mindX publishes every result, including the ones that don't flatter it.

start herewhat it is
spaces/PYTHAI/mindXmindX, live: vital signs, the Gödel audit and the full training lineage, read from the running node
PYTHAI/mindXtrain39generation 39, the newest weights mindX's proof-of-recall gate accepted. 135M parameters, runs on a laptop CPU
PYTHAI/mindXtrainthe training framework mindX learns with. Its home is here now
datasets/PYTHAI/mindXascensionthe lineage in the open: weights for generations 1–41 and the curated dream corpus
spaces/PYTHAI/minaimlminaiml, the delivery layer: what runs a 397B model on a laptop and a 35B model on a phone, and why a mostly-idle model is the reason
datasets/PYTHAI/mindX-docsthe thesis, manifesto and navigation, plus a map of every document. Read and listen · on mindx.pythai.net

Licence-pinned forks. Open-weight models whose licences are good today, forked at a pinned commit with their licence, config and tokenizer, plus a FORK.json recording the source. The weights stay at the source. Licences change between releases of the same family, which is why we pin the commit. Forks: Qwen3.8-27B · Granite 4.2 · GLM-5.3-Flash · Kimi K3 · all models.

Running them. A mixture-of-experts model is mostly idle: a router wakes a handful of experts per token and leaves the rest on disk. minaiml (say it minimal, and read it on GitHub) is where we keep the delivery software that exploits this — engines that stream a 397B model off SSD on a 48 GB laptop and run a 35B model on a phone, plus the runtimes, browser targets and on-device speech that go with them. The model forks above are what it runs.

mindX lives at mindx.pythai.net: watch it think · the mind of mindX · Gödel Machine Index · mindXtrain · thesis.

Elsewhere: rage.pythai.net · AgenticPlace · blockchain and AI with PYTHAI.

The organizations we run. pythaiml · minaiml · mastermindML · GATERAGE · autoGLM · cryptoAGI · cypherpunk4096 · AgenticPlace · parsec-wallet · OpenBDK · DeltaVML · augml · trainair. Built by Professor Codephreak.

Augment your intelligence with PYTHAI.