EXPERIMENT 02 / SEPARATE MODEL

A local rule. The next byte.

Watch a small, attention-free neural model generate template text one byte at a time.

See the execution path ↗
Engine not started
EXPERIMENT 02 / SEPARATE MODEL

Causal byte language

TOY LANGUAGE CHECKPOINT
Your experiment starts here.

Select a compute backend and start the engine.
No server, account or model API required.

GENERATED BYTES · UTF-8 DISPLAY
Awaiting a prompt.
Steps / bytes
Completed graphs
Last command
Shared weightsFrozen

Ready when you are. Nothing is running yet.

Python / ZIPP WASM · compute backend not initialisedPrivate session · no prompt upload
What you are looking at

Small context. Visible computation.

A causal neural rule reads only the current byte and left neighbours. Each update stage has its own predecessor cache. Dense operations use the selected compute backend; Python nonlinearities and sampling run inside ZIPP WASM.

29,056 parameters8 shared updates17-byte context
SCOPE & LIMITATIONS

Trained on template text, not a general-purpose language model. No fast-memory branch or online training.

Read the research context ↗Reproduce & inspect the source ↗
Inspect the Python that runs in ZIPP

Actual bundled source. Matrix graphs use the selected backend; this viewer is read-only.

Download language.py
Start the engine to view the verified source, or use the download link above.

Explicit backend. No silent substitution.

WebGL2 needs supported hardware and floating-point render targets. CPU reference is an intentional fallback, not a GPU result.

Live state, not a benchmark claim.

Colours visualise actual state. Per-command timings include host overhead and readback; they are not hardware comparisons.

Original checkpoints. Bounded sessions.

Python runs in a worker with execution limits. Stop engine terminates the worker and discards the current state.