The Role of Search Algorithms in Text Generation
A large-scale text generation model is capable of calculating the probability for any given sequence of words. Despite this, finding the single most probable 100-word sequence is considered computationally infeasible. Explain the fundamental reason for this infeasibility and describe the primary role of a search algorithm in this context.
0
1
Tags
Ch.5 Inference - 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
Related
Formula for a Candidate Solution Set
A developer is building a text generation system to find the single most probable 20-word summary of a given document. Their proposed method is to generate every possible 20-word sequence, calculate the probability of each one, and then select the sequence with the highest probability. Which statement best analyzes the fundamental limitation of this approach?
Selecting a Generation Strategy for a Real-Time Application
The Role of Search Algorithms in Text Generation
An engineer is developing a system to generate a 10-word sentence. They are comparing two approaches:
- Approach 1: Systematically generate every possible 10-word sequence from the system's vocabulary, calculate a probability score for each, and select the one with the highest score.
- Approach 2: Generate the first word, then, given that first word, generate the most probable second word, and so on, making a locally optimal choice at each of the 10 steps.
Which statement best analyzes the relationship between these two approaches?
Trade-offs of Greedy Search in Inference