Short Answer

Debugging a Text Generation Process

A developer is building a text-generation model. They provide the input sequence of tokens ['The', 'sky', 'is']. The model's first generated token is ['blue']. For the next step, to generate the token that follows 'blue', the developer's code incorrectly uses only the token ['blue'] as the new input for the model. Explain why this approach is flawed and describe the correct input sequence the model should process.

0

1

Updated 2025-10-09

Contributors are:

Who are from:

Tags

Ch.4 Alignment - Foundations of Large Language Models

Foundations of Large Language Models

Foundations of Large Language Models Course

Computing Sciences

Analysis in Bloom's Taxonomy

Cognitive Psychology

Psychology

Social Science

Empirical Science

Science