1.0 KiB
BatchedExecutor Guidance - Classifier Free Guidance / Negative Prompting
This example demonstrates using Classifier Free Guidance (a.k.a. negative prompting) with a custom sampling pipeline. Negative prompting is a way of steering the model output away from certain topics. See the source code here.
Two conversations are created. The guided conversation starts with the prompt that should be completed as shown as the output, for example "my favourite colour is". The guidance conversation contains the negative prompt at the start, for example "I hate the colour red. My favourite colour is". Note that this is a negative prompt, so therefore this guidance will make the model answer as if it likes the colour red.
A custom sampler samples the guidance conversation and uses that output to influence the output of the guided conversation. Once a token is selected both conversations are continued with this token.