Learn Before
Multiple Choice

An AI model generates a step-by-step solution to a complex math problem. The model's final answer is correct. However, upon review, it is discovered that an intermediate calculation step contains a logical error, but a subsequent error coincidentally corrected the mistake, leading to the right final number. If an evaluator's goal is to reward sound reasoning by assessing the validity of each individual step, how would this solution be scored?

0

1

Updated 2025-10-01

Contributors are:

Who are from:

Tags

Ch.3 Prompting - Foundations of Large Language Models

Foundations of Large Language Models

Foundations of Large Language Models Course

Computing Sciences

Ch.5 Inference - Foundations of Large Language Models

Analysis in Bloom's Taxonomy

Cognitive Psychology

Psychology

Social Science

Empirical Science

Science