Learn Before
Sequence Representation for Reward Calculation in RLHF
To generate a single vector representing an entire prompt-response sequence in a reward model, the sequence is processed from left to right using forced decoding. Because language modeling restricts each position to only accessing its left context, the output from the top-most Transformer layer at the first position cannot encapsulate the full sequence. To resolve this, a special symbol, such as , is appended to the end of the sequence. The corresponding output vector from the Transformer layer stack at this final position is then used as the comprehensive representation of the entire sequence.
0
1
Tags
Ch.2 Generative Models - Foundations of Large Language Models
Foundations of Large Language Models
Foundations of Large Language Models Course
Computing Sciences
Related
Sequence Representation for Reward Calculation in RLHF
A team is developing a model to automatically assign a quality score to an AI-generated response. To do this, the model must be given some text as input. Which of the following best explains why the model should be given the original prompt concatenated with the AI's response, instead of just the AI's response alone?
Reward Model Input Preparation
Debugging a Reward Model's Input Formulation
Learn After
Final Reward Score Calculation in RLHF
A team is building a system to evaluate text sequences. They use a model that processes text one token at a time from left to right, where the output for any given token is influenced only by the tokens that came before it. To obtain a single vector that represents an entire input sequence for scoring, which of the following strategies is most appropriate for this type of model?
Reward Model Implementation Analysis
Critique of a Sequence Representation Method