Optimizing a Real-Time Translation Service
Evaluate the lead engineer's proposal described in the case study. What is the primary advantage of this proposed change, and what is the most significant potential drawback for the app's users? Justify your reasoning.
0
1
Tags
Ch.5 Inference - Foundations of Large Language Models
Foundations of Large Language Models
Foundations of Large Language Models Course
Computing Sciences
Evaluation in Bloom's Taxonomy
Cognitive Psychology
Psychology
Social Science
Empirical Science
Science
Related
A company is developing an AI for generating marketing slogans. They are testing two different text generation configurations:
- Configuration A: At each step of generating a slogan, the system instantly picks the single most statistically likely word to come next. This process is very fast.
- Configuration B: At each step, the system considers several different high-probability words, temporarily builds out the different resulting phrases, and then selects the phrase that is most coherent and creative overall. This process is much slower and uses more computational power.
Which statement best analyzes the trade-off between these two configurations?
For a system that generates text one word at a time, the strategy of always choosing the single most statistically likely next word is both the most computationally efficient method and the one that guarantees the most coherent and contextually optimal final output.
Optimizing a Real-Time Translation Service