Skip to content
The Chat Archive
Human-written demonstration (OpenAssistant volunteers)· OpenAssistant (human-written) OpenAssistant Conversations (oasst2)

What is the difference between a generative model and a discriminative model, and what…

In OpenAssistant data, “assistant” turns were written by human volunteers as demonstrations. They are not model outputs.
  1. User

    What is the difference between a generative model and a discriminative model, and what are the trade-offs between each approach in artificial intelligence?

  2. Assistant

    In artificial intelligence, a generative model is one that can generate new data or results based on given inputs. It generates fresh data that is consistent with the training data and the underlying probability distributions of the model. A discriminative model, on the other hand, is one that has been trained on labelled data and can categorise incoming data into one or more predetermined categories. Generative models are better at creating new data, but discriminative models are better at classifying data. The trade-off between the two approaches is that generative models can demand significantly more data and computational capacity, whereas discriminative models are faster and require less data.

  3. User

    Thank you for the information. You are correct that generative models are better at creating new data, but discriminative models are better at classifying data. The trade-off between the two approaches is that generative models can demand significantly more data and computational capacity, whereas discriminative models are faster and require less data.

    Is there anything else I can help you with?

  4. Assistant

    In AI, a generative model is one that can create new and varied information and outcomes based on current data. It processes fresh variables that are computed with binary options and the probability variables of a given template. A discriminative model, on the other hand, is one that has been trained on labelled data and can categorise incoming data into one or more predetermined categories. Generative models are better at creating new data, but discriminative models are better at classifying data. The trade-off between the two approaches is that generative models can demand significantly more data and computational capacity, whereas discriminative models are faster and require less data.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.