Los

Biomedical Engineer doing drug target discovery using both computational and experimental approaches

[Los @Villapudding]

@melvindvivas Codex is good for what I need it to do. And the most important part is that it doesn’t constantly hit guardrails in bio work

[Los @Villapudding]

@Sathclark @p_e_cooper To add to this, if you fail they overrun the DJ booth and you die

[Los @Villapudding]

@MozeTech This is actually insane. Although, when you first destroyed that tree thought a pidgey or something would spawn in lol

[Los @Villapudding]

@rynorhn And they lived happily ever after

[Los @Villapudding]

@QuanquanGu @Geodesiclab This is some exciting stuff. What kind of drug discovery do you do Dr. Gu?

[Los @Villapudding]

@mosasaurus27 Why is one thing as others have pointed out, the bigger problem is simply cost. It’s expensive to run some of these experiments.

[Los @Villapudding]

@shengzheyao Idk about antigravity. At least the desktop app. I loved it at first and then eventually it became agents centered and less of an IDE interface and eventually just returned to VS Code. It was a bummer

[Los @Villapudding]

@RandomOhioCards @roywoodjr On god. Easy

[Los @Villapudding]

Im looking forward to testing Argon. Gemini pro was a life saver when discussing experiments and quickly catching mistakes, GPT-Sol 6.1 has been fantastic so far as well, but it’d be nice to see another frontier model where I can actually brainstorm biomedical research.
Opus and Sonnet just flag almost anything research related even if it’s not anything crazy (Flow Cytometry for example)

[Los @Villapudding]

@SynBio1 Yall not able to afford a vortexer?

[Los @Villapudding]

@rovvmut_ This looks pretty accurate to Dario tho

[Los @Villapudding]

Interesting to see some circle jerk going on about Opus and Sonnet. They’re good models yeah, but I think in most realistic scenarios you don’t need to be on the bleeding edge for everything. I keep on coming back to Codex for one reason: it doesn’t immediately trigger guardrails all the time like both Opus and Sonnet.

Something I just need to discuss an experimental plan but simply can’t with Anthropic’s models.

[Los @Villapudding]

@adibvafa This is what I also noticed. Just even discussing some flow cytometry data immediately triggered the guardrails. As a researcher, it’s unusable for biomedical research…

[Los @Villapudding]

Mmmm are we really thinking this is because of safety? Or does Astra not live up to compete against Sonnet and Opus?

[Los @Villapudding]

@FNFL_Historique The Putin propaganda is getting pretty crazy

[Los @Villapudding]

@ymmmayer Ability to travel between higher ordered dimensions.

[Los @Villapudding]

@nickmmark Wait people thought this actually did something? Lmao

[Los @Villapudding]

@philosophymeme0 This is me as I’m trying to get all my experiments done just to get to the Sunday evening and start planning my week again

[Los @Villapudding]

@simonmaechling Yet people don’t understand this and it’s unfortunate…

[Los @Villapudding]

@Google @GoogleResearch Is there a way to run this in a simulation and conditionally train the fly?