Start learning →

Rebuild the model width from heads

Exercise and reference answers

Sequence
  1. Item
    64 dims
  2. Item
    128 dims
  3. Item
    256 dims
  4. Item
    512 dims
Instructions
Drag one head width onto an equal head width until the eight heads add back to full width.

Explanation

_Source: Qallery verified question bank, AI Foundations, topic transformer_attention_
Continue learning with Quill →

Browse more Claude questions and exercises →