Google

Once a process-based reward model is trained to classify each step of a reasoning path, it can be used to evaluate the overall quality of the entire path. A straightforward evaluation method involves aggregating the step-level classifications by simply counting the total number of steps that the model identifies as 'correct'. This count serves as the overall score or reward for the reasoning path.

Scoring Reasoning Paths by Counting Correct Steps

A simple method to score a reasoning path is to count the number of steps classified as 'correct'. This can be represented by the formula: 

$$r(\mathbf{x}, \mathbf{y}) = \sum_{k=1}^{n_s} \delta(\text{correct}, C(\mathbf{x}, \bar{\mathbf{y}}_{\le k}))$$

where:
- $$r(\mathbf{x}, \mathbf{y})$$ is the total reward for the reasoning path $$\mathbf{y}$$ given the input $$\mathbf{x}$$.
- $$n_s$$ is the total number of steps in the reasoning path.
- $$C(\mathbf{x}, \bar{\mathbf{y}}_{\le k})$$ is the classification output for step $$k$$, determined by selecting the label with the maximum probability.
- $$\delta(a, b)$$ is the Kronecker delta function, which is $${}1$$ if $$a=b$$ and $${}0$$ otherwise. In this context, it equals $${}1$$ if the classification for step $$k$$ is 'correct'.

Formula for Scoring Reasoning Paths by Counting Correct Steps

Imagine a method for evaluating a multi-step solution generated by an AI. In this method, the overall quality of the entire solution is determined by simply counting the total number of individual steps that are identified as 'correct'. Given the case study below, calculate the final score for the AI's reasoning path and briefly explain how you arrived at your answer.

Scoring an AI's Reasoning Process

An AI model generates a 5-step reasoning path to solve a problem. A separate classification model evaluates each step and produces the following labels: [Correct, Correct, Incorrect, Correct, Incorrect]. Based on the method of scoring by counting the number of correct steps, what is the final score for this reasoning path? Briefly explain how you arrived at this score.

Calculating a Reasoning Path Score

An AI's multi-step solution to a complex problem is evaluated by a separate model that classifies each step as either 'correct' or 'incorrect'. The final quality score for the entire solution is calculated by summing the total number of steps classified as 'correct'. What is a primary conceptual limitation of this evaluation approach?

Learn Before

Related