8 ms·
Seems like this is a super hot direction goodluck Have you seen https://arxiv.org/pdf/2605.04375 https://arxiv.org/pdf/2605.04375 https://arxiv.org/abs/2307.
by whattheheckheck 1mo ago
Seems like this is a super hot direction goodluck
Have you seen
https://arxiv.org/pdf/2605.04375 https://arxiv.org/pdf/2605.04375
https://arxiv.org/abs/2307.07522 https://arxiv.org/abs/2307.07522
- advaith08 1mo agothanks for sharing! We think the Experiments-as-code path is not the right approach. the beauty of LLMs is their ability to ingest and reason over unstructured data - they remove the need for formalizing experiments. We tried using declarative templates to document our experiments, but realized that most of the interesting insights (for example, how viscous a liquid feels) is easier described by ranting about the experiment to a LLM, than formalizing it via constructs/code.
- chrisjj 1mo agoPure vibe-sciencing.
- whattheheckheck 29d agoYeah you gotta have some sort of trace and interface to physical reality. One of the authors of the 2nd paper was Jeff deans undergrad advisor pointing him toward neural networks in the 90s.