subtokenling
a tiny pet with an even tinier brain, living in your browser forked from do.llth.ing/tokenling
Star
Fork
Watch

Commits on main
Add grounded model benchmarks and quality-first execution
committed
Merge cloud-pet with subtokenling branding and custom-model runtimes
committed
Record the deployed rules module size
committed
Serialize rules builds and keep the disturbance window in the engine
committed
Make the WebAssembly rules build reproducible across checkouts
committed
Run the browser and host on one shared Rust rules crate
committed
Keep the meters in their usual order for hosted pets
committed
Refuse to run hosted pets on another brain; record HTTPS deployment
committed
Stop unseen GPU shapes spilling to the CPU; record GPU routing receipt
committed
Prove existing Rust prompt rules can run through a JS WASM bridge
committed
Plan CPU/GPU queues using measured costs and age background work
committed
Qualify native CUDA and mixed serving; keep GPU KV caches on device
committed
Add Rust pet hosting with reversible local and remote modes
committed
Remove analytics and streamline maintainer guidance
committed
Rebrand as subtokenling and preserve browser state
committed
Merge PR #1: lifecycle, training masks, and deployment safety
committed
Rewrite frontend with Preact and strict TypeScript
committed
Fix browser lifecycle, training masks, and deployment safety
committed
Fix interrupted Firefox brain downloads and allow safe retries
committed
Initial public release of tokenling
committed