Learn Before
Formula for the Predictive Distribution in Bayesian Prompt Ensembling
In the Bayesian framework for prompt ensembling, the predictive distribution of an output given a problem is calculated by marginalizing over the prompt . This integral computes the total probability of by considering all possible prompts, weighted by their respective likelihoods. The formula is expressed as: . Here, is the predictive distribution of the output given a specific prompt, and is the prior distribution reflecting the probability of the prompt given the problem.
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
Formula for the Predictive Distribution in Bayesian Prompt Ensembling
Robustness of the Bayesian Prompt Ensembling Model
An AI development team observes that their model's performance on a specific problem is highly dependent on the exact phrasing of the input prompt. Their current strategy involves testing a small, fixed set of prompts and aggregating the outputs. To build a more fundamentally robust system that is less sensitive to these variations, which of the following represents the most effective conceptual shift in their approach?
Conceptual Shift in Prompt Handling
According to the Bayesian view of prompt ensembling, the process is fundamentally about identifying the single best prompt that maximizes the likelihood of the desired output for a given problem.
Uniform Prior Assumption in NLP Prompting
Learn After
Computational Infeasibility of the Bayesian Predictive Distribution Integral
A team is using a probabilistic method to combine the outputs from a language model for a variety of different prompts (x) to solve a single problem (p). The final probability of a specific output (y) is calculated by integrating over all possible prompts. The formula for this is: Pr(y|p) = ∫ Pr(y|x) Pr(x|p) dx. In this formula, Pr(y|x) is the model's likelihood of the output given a prompt, and Pr(x|p) is a prior distribution representing the assumed suitability of a prompt for the problem. How would the calculation of Pr(y|p) be affected if the prior distribution Pr(x|p) was assumed to be uniform, meaning every possible prompt is considered equally suitable?
Calculating Predictive Probability with Prompt Priors
A probabilistic approach to combining outputs from different prompts for a single problem involves the following formula: Match each mathematical term from the formula with its correct conceptual description.