Combination Strategies for Prompt Ensembling
The Combine(·) function, used to aggregate multiple predictions in prompt ensembling, can be implemented through various strategies. Examples include non-probabilistic methods like selecting the final output via voting among the different predictions, or by choosing the prediction that shares the most content or overlap with the other candidates.
0
1
Tags
Ch.3 Prompting - Foundations of Large Language Models
Foundations of Large Language Models
Foundations of Large Language Models Course
Computing Sciences
Related
Combination Strategies for Prompt Ensembling
A team is developing a system to generate a single, highly accurate description for an image. They provide a language model with the same image but use five different instructional phrases to guide the description generation process. This results in five distinct text descriptions of the image. The team's final step is to aggregate these five descriptions into one consolidated output. How does the general formula for combining multiple predictions,
ŷ = Combine(ŷ₁, ..., ŷₖ), map to this team's process?Analyzing a Failure in Prediction Combination
Deconstructing the Prediction Combination Formula
Learn After
A developer is using a technique to improve the reliability of a language model's output. For a single request to summarize a text about cellular biology, the model generates the four different one-sentence summaries listed below. After processing these four candidates, the system selects 'Summary A' as the final output.
- Summary A: The mitochondrion is the cell's power plant, generating ATP through cellular respiration.
- Summary B: Often called the powerhouse of the cell, the mitochondrion produces the energy currency, ATP.
- Summary C: The primary role of the ribosome is to synthesize proteins for the cell.
- Summary D: The mitochondrion, known as the cell's powerhouse, is responsible for creating ATP.
Based on the provided summaries and the final selected output, which combination method was most likely used to aggregate the candidates?
Choosing an Aggregation Strategy for Financial Data Extraction
Applying Combination Strategies in Prompt Ensembling