Learn Before
Output Stage in 1-Best Selection
The output stage, labeled as step ③, is the final step in the 1-best selection process. Following the ranking of candidates, this stage involves selecting the single candidate with the highest probability as the final output. All other lower-ranked candidates are discarded, or 'pruned'. For instance, after ranking, 'cute' is selected as the 1-best output, while the remaining candidates ('on', 'sick', 'are', '.') are pruned.
0
1
Tags
Ch.5 Inference - Foundations of Large Language Models
Foundations of Large Language Models
Foundations of Large Language Models Course
Computing Sciences
Related
Example of 1-Best Selection
Ranking Stage in 1-Best Selection
Expansion Stage in 1-Best Selection
Output Stage in 1-Best Selection
Predicting the Next Word
A language model is determining the next word in a sequence. It follows a process where it first creates a list of possible words, then organizes them by likelihood, and finally chooses the most probable one. Arrange the formal stages of this process in the correct chronological order.
A language model is using a three-stage process (Expansion, Ranking, Output) to select the next word for the phrase 'The cat is...'. The model first expands the possibilities to a set of candidates with their probabilities: 'sleeping' (0.5), 'cute' (0.3), 'on' (0.15), and 'blue' (0.05). However, the model's final output is the word 'on'. Which stage of the process is the most direct point of failure?
Learn After
A system for generating text has produced a ranked list of potential next words and their associated scores:
{'the': 0.52, 'a': 0.21, 'one': 0.15, 'that': 0.08}. If the system's final step is to select only the single highest-scoring option and discard all others, what will be the output?A text generation system follows a procedure where, after ranking potential next words by their scores, it selects only the single word with the highest score and permanently discards all other options. What is the most significant potential drawback of this final selection method?
Consider a text generation system where the final step involves selecting the next word from a ranked list of candidates. If this final step's procedure is to choose only the single most probable option, it will select the top candidate and discard all others, even if the second-ranked candidate has a very high probability score.