Good: I hired an awesome DE and a DS research chemist!!
Sad: all those lovely ppl whom I had to reject! Including about 300 who applied late enough that with my capacity (one me, no HR) I never even had the chance to consider them properly! (the screening interview was 30 min, technical one - 2 h)
Im like 90% sure they’re about to make some kind of commercial play in biotech
Fable update:
I needed to select a well-scoped research subproject from a very complex large project, min-maxing both the internal consistency and the presentability of said subproject.
Sonnet just could not even get it
Opus could, with very slow, step by step
Fable on-shot for 10% of weekly budget
Ok so basically I'll have to wait half a year till open models catch up. That's ok. Unpleasant but ok.
Homeostatic dimensional degeneracy during development strikes again!
I wonder if, in a way, it make sense to learn Claudish, not the other way around. Not sure I can imitate yet (lots of lingo; not just poetic: a limited argot of functional poetry), but may be genuinely useful if most models become de-facto optimized for this way of expressing ideas.
Huh wow so I was just lucky that I asked fable about flammable explosive chemicals, chromium vi compound, and the use of RDkit (🌿not flagged🌿) and not about the the powerhouse of the cell (🚨flagged🚨)
This is really cool and also the traffic jam cases are almost better, the delicate feathering structures. I love traffic
math.mino.mobi/traffic
