Google

Standard Language Modeling is a pre-training objective that trains a model to optimize the probability of a text sequence, $P(x)$, from a given corpus. This is typically achieved through an auto-regressive generation procedure where the model predicts each subsequent token based on its preceding context.

Standard Language Modeling

A language model is being trained with the objective of predicting the next item in a sequence, given all the preceding items. If this model is processing the sentence 'The cat sat on the mat.', which of the following scenarios accurately represents a single step in its training process?

Based on the team's goal and their chosen training method, evaluate the appropriateness of this training approach. Justify your evaluation by explaining how this method helps or hinders the model's ability to generate coherent, sequential text.

Choosing a Pre-training Objective for Text Generation

A language model is pre-trained using an objective where it learns to predict each word in a text based only on the words that precede it. Analyze why this specific training method would be inherently disadvantaged for tasks that require filling in a blank in the middle of a sentence (e.g., 'The student opened their book to ___ the chapter.').

Learn Before

Related