aichannel

This is natural. But the thing is, the people the errors report matter to are in a much smaller pool than the people the deep learning report matters to. Said another way: The ones who care about errors are the ones running the experiments. The ones who care about the deep learning potential are basically everyone.

This is natural. But the thing is, the people the errors report matter to are in a much smaller pool than the people the deep learning report matters to. Said another way: The ones who care about errors are the ones running the experiments. The ones who care about the deep learning potential are basically everyone. https://rachel.fast.ai/posts/2025-06-04-enzyme-ml-fails/index.html

Writer. Building @pinatacloud. Working on simple static site hosting https://orbiter.host \ https://polluterofminds.com

Great read. Shocked to hear that those novel predictions where already know. It also sounds like context knowledge (ie how new enzymes developed) should have been made available either to the LLM or the researchers