halfeatenscone

It only takes a basic understanding of neural nets to recognize this as crankery. Giving an LLM a tricky question does not force it to do any more computation than a simple prompt. There is no conceivable way an especially difficult prompt would make the model crash.

I don't think this follows. Calling it a derivative work already feels like a stretch, but even granting that framing, the use is clearly transformative and therefore likely to be considered fair use in the US.

I actually wrote a Wikipedia article on the intersection of copyright law and deep learning models the other day (https://en.wikipedia.org/wiki/Artificial_intelligence_and_co...). I was hoping to include a section on the copyrightability of model weights, but was sadly able to find 0 coverage in reliable sources.

Such logs would not be used for training the base model, but rather for fine-tuning the model for instruction following. Instruction tuning requires far less data than is needed for pre-training the foundation model. Stanford Alpaca showed surprisingly strong results from fine-tuning Meta's LLaMA model on just 52k ChatGPT-esque interactions (https://crfm.stanford.edu/2023/03/13/alpaca.html).

"Classify the following strings according to whether they represent Python programs which terminate..."

I'm not sure how easy it is to grok these by playing with them. If you're confused, the "About" section has some explanation of how the matrices are constructed and how to interpret them: https://colinmorris.github.io/SongSim/#/about

Also, in case it's not clear, you can press the "+" button next to the song dropdown to open a textarea where you can paste arbitrary lyrics.

It's guaranteed to not exist in the training set (i.e. to not be a real GitHub repo that existed before 2015). But the model can certainly repeat itself.