robot.txt doesnt really mean anything, I used to work for a company that scraped the web and this was literally not a concern. That being said, using data for training LLMs is a new things and potential lawsuits going reddit's way are a possiblity, we can't really know.
One note, its name is not gpt-2 it is gpt2 which could indicate its a "second version" of the previous gpt architecture, gpt-3, gpt-4 being gpt1-3, gpt1-4. I am just speculating and am not an expert whatsoever this could be total bullshit.
One note, its name is not gpt-2 it is gpt2 which could indicate its a "second version" of the previous gpt architecture, gpt-3, gpt-4 being gpt1-3, gpt1-4. I am just speculating and am not an expert whatsoever this could be total bullshit.