Re:Gemini 4 Argon
by orbital-decay on Wednesday October 14, 2026 @08:45AM
(#50086865)
Perhaps you're right, this would probably take some rigorous forensics to resolve definitely. Somehow all models have very similar patterns in text, seemingly at random. I don't think everyone is trying to train on specific patterns on purpose, it's a pretty general phenomenon. (I'm talking about mode collapse, not model)
> That's the result of training on the same datasets and scraped web contaminated with other models' outputs
Maybe I'm misinterpreting though, or just semantics. I would consider contamination, or whatever the first example was that they all matched, as specific training data, along with all the duplications of. I doubt it's model collapse to the same pelican, without the same original example.