The Midjourney machine costs <1/10th of an MRI machine and the scans are done in a minute (vs an hour for a whole body MRI).
Should make the scans near-free, easy to do regularly for longitudinal study.
Different use case than an MRI
@dr_alphalyrae I was wondering why all my sessions today were nail-biting over things I do all the time and it turned out I was on auto mode instead of this. Never again!
FIFA is like literally don't believe your eyes instead trust an organization that thought it was a good idea to hold a World Cup in December in fucking Qatar
@furman I was starting another biofx rust rewrite this week (.9999 concordance and 50x better performance with my first one). Was so excited to get Fable to work on it but…
Idea: don’t brag about your model’s capabilities in biology and then be surprised by everyone in biology trying it out and getting pissed. How is this reaction a surprise?
I know this is frustrating. As we said at launch, Fable blocks bio requests entirely for now and reroutes them to Opus 4.8. We made this tradeoff in order to get the model out safely and quickly while we work to refine the classifiers. It is absolutely our goal to enable the whole bio community to use our most powerful models, with appropriately scoped protections in place. We’re working towards this as fast as possible.
@dagarfield@cremieuxrecueil Crashed out on me when I asked it to review a tech spec for a VCF ingestion pipeline update. Tons of excitement for this inside Natera, real buzzkill seeing it’s effectively useless for all our biofx workloads.
Just wrapping up porting a major biofx tool from Perl to Rust with LLMs. Worth it (a few months as a side project) but the complexity of DNA splicing made them cry on multiple occasions, even with the right answer to reference.
I think the zig to rust port with an LLM is one of the more ideal uses of an LLM. I assume it would excel at this for its literally the purpose of an LLM.