Pinned
Chaitanya K. Joshi
@chaitjo
AI for biology 🧬
Postdoc at Stanford @RDasLab
PhD @Cambridge_Uni · prev @AIatMeta @PrescientDesign @MRC_LMB
Feel kind of sad. Obviously it was said that as soon as benchmarks go public the labs train on the test set or RL on them. But this reminds me of school, rote learning for doing well in exams, etc.Replying to @kotekjedi_mlBut we also took a chance to have a look at some in-the-wild scheming, reward seeking, etc. examples, and dumped it in appendix. 1) Summarizer unfaithfulness Reasoning summaries often omit important information from the original trace. Here, Opus 4.8 realizes it knows the- Sometimes in academia ppl will say stuff like chai, protenix, boltz, openfold, rf3, etc. are all AF3 'clones' with an undertone around novelty, etc. I think this is such a wrong way to look at it, these models are all foundations for real products now and also importantly the"it's compounding returns. every week making 5 to 10% improvements - stack those together for a long time and you end up in a very different place than where you started.” -@joshim5
- Maybe the ribosomal proteins are like the harness around the ribosomal RNA which is the intelligenceThe ribosome is an RNA you absolute moron
