victor casas

Head of Data and AI - Barclays

[victor casas @vittycasas]

@abemurray Hi

[victor casas @vittycasas]

@sprmn2024 @NousResearch Congrats!

[victor casas @vittycasas]

@Steve_Yegge Happy birthday my friend, you are a true and incredible inspiration!

[victor casas @vittycasas]

@systematicls Of course, that’s how boosting and bagging work. That’s why the xgboost has been such a beast

[victor casas @vittycasas]

@Hesamation I love this one honestly! Finally some sense in reasoning outsourcing!

[victor casas @vittycasas]

@systematicls Interesting as always!

[victor casas @vittycasas]

@systematicls You are always on point! Love the hints

[victor casas @vittycasas]

@ashpreetbedi Your memory is unbounded, so more interactions per user then the system becomes biased to the power users. Additionally, how do you maintain knowledge relevance? There is a decay factor in knowledge too.

[victor casas @vittycasas]

@systematicls I thought this was very well known.

[victor casas @vittycasas]

@dejavucoder Thank you!

[victor casas @vittycasas]

@oznova_ Put me in coach!

[victor casas @vittycasas]

@investingidiocy 10 seats please!

[victor casas @vittycasas]

@__paleologo @Aion19834333 Remember the claude code is a CLI of the Anthropic models, what codes add is the Bash loop and connectivity to other services, but it is all the Anthropic family behind the wheels. You can do CC with either Haiku, Opus or Sonnet if you want an independent view of them.

[victor casas @vittycasas]

@__paleologo @amendandpretend This is the true alpha.

[victor casas @vittycasas]

@planert41 Yep, count me in boss.

[victor casas @vittycasas]

@__paleologo Tab has been around for a while, but it only works for small datasets, lets see what this new version has then.

[victor casas @vittycasas]

@predict_addict @macrocephalopod I can tell you from personal experience this is not true.

[victor casas @vittycasas]

@spwfeijen Nano

[victor casas @vittycasas]

@NoahEpstein_ Interested!

[victor casas @vittycasas]

@elonmusk Yes!