← All workshopsRead Autoregressive Generation

Large Language Models

Return the prediction to the input

BOTIE

Use the fixed next-token table to predict from the last context ID. Choose the highest-scoring token and append it to the context. The next example begins with a different context.
Ready to test
0 passed · 0 failed · 1 not run

An empty floor. Start with the information your machine needs.

Received result

Your answer will arrive here.

Examples, experiments and workspace controls
Live values

Connect a tool's inputs to see what it produces. You can try a branch before connecting it to Result.

Try your own inputs

Use a copy of the example to try your own data. Keep the problem's assumptions in mind, and predict the effect of a change before checking the output. You can also add another branch to compare two ways of doing the calculation.

Start at an output arrow, then click or drag to an input. Drag empty space to pan. Delete removes a selection. Ctrl/Cmd+Z undoes a change.
Context for this challenge

Use the fixed next-token table to predict from the last context ID. Choose the highest-scoring token and append it to the context. The next example begins with a different context.

Botie has predicted one token. Continuing the sequence requires returning that choice to the next input.

Use the fixed table with the last context ID, choose its highest score, and append the selected ID.

Examples and results
Current context IDs
  1. 00
Fixed next-token score table
031
004
500
Required result
  1. 00
  2. 11
Try a fresh construction

This clears this draft, including its undo history. Earlier successful attempts stay recorded. Build again without opening hints to record an independent attempt.

Stuck? Inspect a worked construction

The tools below form one possible solution. You can assemble it, follow its signals, and undo that change to return to your own construction.

Assumptions and test precision

This builds one greedy generation step with a tiny fixed model. Repeated generation feeds the growing context back through the same model; its weights do not update. Numerical tests allow a difference of 0.001. Lists and structures must also have the right shape.

The toy generator loops back

This check uses the pieces from “Append the greedy token to the context”.

BOTIE

Botie's context already contains IDs 0, 1 and 2. The toy model chooses its score row from the last ID in that context. Make one greedy generation step and return the full extended context.
Current context IDs
  1. 00
  2. 11
  3. 22
Fixed next-token score table
031
004
500

Use a number, text, or an array matching the requested result. For a matrix, put each row in brackets. Saving your prediction runs your current construction on these inputs and checks the earlier examples too.

Build a construction in the workspace above first. This case will test those pieces.

Which state changes between generation steps if the model weights stay fixed?

An autoregressive loop updates its context; ordinary generation does not train the model.

This builds one greedy generation step with a tiny fixed model. Repeated generation feeds the growing context back through the same model; its weights do not update.

Work through the Python challenges →
Continue the courseSampling and Temperature →
← Next-Token Prediction