Heading to San Diego for #NeurIPS2025 (Dec 1–9).

Happy to connect with people working on transfer & generalization, time-series/EEG foundation models, or biotech.

How Deepseek can help with bandwidth?

That's an argument for 'small efficient training is possible', not against 'more GPU means better product [anyways]'.
"less demand for GPU" is not trivial.

Does "biblical" mean a story we wish were true although we don't have any concrete reason to believe it will actually happen?

🙋‍♀️