Critique of a Training Method for a Story-Writing AI
An AI development team is training a model to write short, creative stories. Their training method involves having the model generate a complete story, and then a human evaluator gives the entire story a single score from 1 to 10. The model is rewarded based on this single, final score. After extensive training, the team finds that while the model's grammar has improved, its ability to create a coherent plot with logical character development remains poor. Explain the fundamental reason why this training approach is ineffective for teaching the model complex skills like plot construction.
0
1
Tags
Ch.4 Alignment - 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
Related
Aspect-Based Sentiment Analysis as an Example of Granular Evaluation
Segment-Based Reward Computation
Importance of Step-by-Step Supervision for Complex LLM Reasoning Tasks
Debugging Common C Syntax Errors: A 'Hello, World!' Example
Example of Outcome-Based Reward for a Mathematical Task
A research team is fine-tuning a language model on two different tasks. For which of the following tasks would a reward system that only provides a single score based on the final output's correctness be the least effective for identifying and correcting errors in the model's generation process?
LLMs for Textual Error Correction
Diagnosing a Flawed LLM Training Strategy
Critique of a Training Method for a Story-Writing AI
Aspect-Based Sentiment Analysis (ABSA)
Process-Based Supervision for Complex Reasoning