A text generation system is configured to keep the top 3 most probable words at each step. Given a list of five candidate words with their probabilities, arrange the following actions in the correct sequence to determine which words to keep.
0
1
Tags
Ch.5 Inference - Foundations of Large Language Models
Foundations of Large Language Models
Foundations of Large Language Models Course
Computing Sciences
Comprehension in Revised Bloom's Taxonomy
Cognitive Psychology
Psychology
Social Science
Empirical Science
Science
Related
A text generation system is deciding on the next word in a sentence. It has calculated the probabilities for five potential words. If the system is configured to keep only the top 3 most probable options for the next step, which set of words will it select? The potential words and their probabilities are: 'the' (0.45), 'a' (0.25), 'his' (0.15), 'her' (0.10), and 'its' (0.05).
A language generation model is configured with a beam width of 3. At one step, it considers the following five words and their associated probabilities: 'happy' (0.40), 'joyful' (0.25), 'glad' (0.18), 'content' (0.12), 'pleased' (0.05). Which words will be pruned (discarded) during this selection step?
A text generation system is configured to keep the top 3 most probable words at each step. Given a list of five candidate words with their probabilities, arrange the following actions in the correct sequence to determine which words to keep.