Learn Before
Using Optimized Predictions as Learning Targets
In certain training methodologies, the target for the learning process is generated by the model itself rather than being a pre-existing ground truth label. This involves identifying the output that maximizes an objective function, such as a log-probability. This optimized output is then used as the target for a loss function, which in turn guides the model's parameter updates.
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
Using Optimized Predictions as Learning Targets
General Formula for Prediction via Maximum Probability
A language model is given the input 'The ocean is...' and calculates the conditional probability for four candidate words to be the next word. Based on the values below, which word would a model that predicts by maximizing probability choose?
- P("deep" | "The ocean is...") = 0.75
- P("cold" | "The ocean is...") = 0.15
- P("running" | "The ocean is...") = 0.09
- P("quiet" | "The ocean is...") = 0.01
Evaluating a Prediction Strategy
Spam Filter Classification
Learn After
Log-Probability Loss with Model-Generated Target
A research team is training a generative model using a method where the learning target for any given input is the output that the model itself currently calculates as having the highest probability. This self-generated target is then used to update the model's parameters. Which statement best analyzes a key implication of this training approach?
Self-Reinforcing Training Strategy for a Chatbot
Contrasting Learning Target Methodologies