MacroStories by raincandy-u ↗ Original model ↗

A continuation model, not a chatbot. Try a story opening. Alex, Robin and Casey map to its character tokens.

Empty opening · starts from the story token.

Make it your experiment
0.50
0.90

Leave the seed empty or enter -1 for a new random seed each run. A fixed seed makes the experiment repeatable.

Preparing Transformers.js…

Every little story
starts somewhere.

Choose an opening, or leave it to chance.
Then see what 20K parameters can do.

WORDS—
NEW TOKENS—
TOKENS / SEC—
TIME TO FIRST—
2,048-token context · FP32 inference
Look under the hood Raw text & token IDs

Character markers are displayed as names above. Here is the unmodified decoded output, including any unknown tokens.

No output yet.
[]