Vasco Gameiro

Devagar 🫳

[K @iiamkrshn]

Congratulations on working 40 hours! You are now free to dissociate for 2 days

[ksa 🏴‍☠️ @kosa12m]

How Anthropic talks about Claude Mythos rn:

[yagi @yducknow]

how it feels to speak during a work meeting when u are under 30

[Anime Shots @anime_shots]

❤️‍🩹

[‏ً @ewedunega]

my favorite hero doesn't wear a cape, he's poor, bitter and stalks his ex ❤️

[kanav @kanavtwt]

Someone built a Google translate for Linkedin 😭

[Rob Hamilton 🟥 @Rob1Ham]

IBM in 1979:

[Vasco Gameiro @vascomgameiro]

@predict_addict 2. My current default stack is Polars for dataframes, GBTs for most modeling, Optuna for hyperparameter optimization, scikit-learn for preprocessing , metrics and some other models, Nixtla for forecasting and rapids when I have access to gpu. What alternatives would u recommend?

[Vasco Gameiro @vascomgameiro]

@predict_addict 1. For local dataframes, I agree pandas is showing its age — Polars, DuckDB, and cuDF (with GPU) are better choices. That said, I still haven’t found a great replacement for local tabular ML development: scikit-learn is still hard to beat for preprocessing and metrics on CPU.

[westerosies @westerosies]

HBO posted this in response to the ‘A Knight of the Seven Kingdoms’ review bombing from Breaking Bad fans ☠️😭

[Game of Thrones @GameOfThrones]

The realm needs good men.