tinyfiles.tv
tinyfiles.tv β€” Meet Roger

Meet Roger.
A state-of-the-today
learning model.

Roger is learning. Right now. Today. He is not smart yet. He needs your help to improve.

πŸ‘€
ROGER
😐
v0.0001 β€” today model
Roger says:

Hello. I am Roger. I think 1+1 is 11? And blue is... red? Please help me. I want to be smart.

Roger Brain: 2% β€” Tasks done: 0 / 4 β€” Mood: confused

Teach Roger β€” 4 braindead tasks

If you can do these, you are smarter than Roger.

That is not a high bar. That is the point.

Task 1 / 4 β€” Math for Roger

Roger thinks 1+1 = 11. Correct him.

Click the right answer. Roger is watching.

Task 2 / 4 β€” Colors for Roger

Roger is colorblind (in his brain).

What color is the square? Roger guessed β€œyellow” for all of them.

Task 3 / 4 β€” Drag the ball to the hole

Roger cannot do this. Show him.

Drag the orange ball into the black hole. On phone: drag with your finger.

HOLE
πŸ€

Roger: β€œBall go... where?”

Task 4 / 4 β€” Spelling for Roger

Roger spelled all of these. Are ANY spelled correctly?

Click the words that ARE spelled correctly. Then check with Roger.

Roger: β€œI think speling is correct. It looks right to me.”

The backend, as planned

What would sit behind Roger.

Roger isn't a real model today. The replies, the brain meter and the four tasks above are all scripted in this page and run entirely in your browser; nothing you do here is sent anywhere or used to train anything. The plan is to put a lightweight language model behind him later, small enough to run on modest hardware. The files below are a sketch of what that backend could look like: some are drafts, some are empty placeholders, and all of it will change once a model is actually chosen.