21 ms·
It is zero shot. The llm is trained to generate next token. Zero shot is defined as being able to output predictions for classes they were not trained on. It
by jerrygenser 3y ago
It is zero shot. The llm is trained to generate next token.
Zero shot is defined as being able to output predictions for classes they were not trained on.
It doesn't mean the input data can't be in ml task domain but that the model was not trained on this particular ML task and/or classes.
- mistymountains 3y agoThe training classes in this case are words in the vocabulary in the context of a sentence.
- godelski 3y ago> It is zero shot. The llm is trained to generate next token. I'm going to refer you to the sibling comments as they stated similar things and I answered them in depth and do not wish to repeat myself. But to summarize: Zero-shot := Goal of f:X → Z but train f':X → Y, where Y ⊂ Z. We test on A⊂Z, where A⊄Y (sometimes definition is A ∩ Y = {∅}, but I'm not being as strict) > but that the model was not trained on this particular ML task and/or classes. I'm going to need explicit clarification as to this. Explicitly or implicitly? See sibling comments and note about likelihood and density estimators w.r.t. classification.