Javier Abellán

Digital daredevil.

[Javier Abellán @javier_a2]

@javilop Oye , y habéis probado imágenes sintéticas de textos? Text rendering, etc ..

[Javier Abellán @javier_a2]

@javilop Debe ser brutal el dataset de textos con su blur como data augmentation para entrenar semejante upscaler (y todo made in murcia imagino)👏👏👏

[Javier Abellán @javier_a2]

@antor Llevaba unos años así, escribía muchos .md pero luego me daba pereza leerlos (code editor) ahora estoy con notion más feliz que una perdiz.

[Javier Abellán @javier_a2]

@javisantana Demasiados cracks juntos

[Javier Abellán @javier_a2]

@iScienceLuvr Yes, vision model for different domains: medical image + satellite imagery + wildness imagery + etc.

[Javier Abellán @javier_a2]

@andfkdev He is just advertising his product. He has to say that, and we have to ignore it.

[Javier Abellán @javier_a2]

"You just need harder problems to solve"

This is beautiful quote form @Sentdex . Don't feel pessimistic about AI being smarter than you. Feel optimistic because you can solve harder problems.

[Harrison Kinsley @Sentdex]

I remember trying to force myself to use C++ to do the same things I could do 100x faster in Python. I just couldn't do it. Not because it was hard, because there was no point for that particular problem. There are obviously still many important and useful usecases for C++ today, but the needs have changed over time.

Abstraction is a good and natural thing. It's a core human capability that has given us our entire civilization and allows us to progress technologically.

When assembly hit, the machine coders disliked it. When C++ came around, the C/Assembly people disliked it. C++ people still often dislike Python. It's definitely okay to be sad about change, and Mo's take here seems super human and genuine, recognizing mostly that times are changing, but the argument is false.

Mo is not useless. Mo just needs harder problems to solve.

[Andrés Miguel Torrubia Sáez @antor]

AI deniers are starting to sound like flat earthers.

[Javier Abellán @javier_a2]

@javilop @psluaces Con tantos posts, como encuentras posts antiguos con tanta facilidad? Haces scroll en tu perfil o haces magia negra con algún tipo de semantic vector search ?

[Javier Abellán @javier_a2]

@naroh cc @JaimeObregon

[Javier Abellán @javier_a2]

@nico_dufour @AIatMeta @p_bojanowski @VickyKalogeiton @david_picard @loiclandrieu Hi @nico_dufour, is this PLONK model (DINO v3 based) available ? I cant't find it on github neither HuggingFace. It will be awesome if this model is public :)

[Javier Abellán @javier_a2]

@mrm8488 Mmm parece una técnica similar a VL JEPA no? Parece que es tendencia representaciones semánticas intermedias

[Javier Abellán @javier_a2]

@tunguz Russians will know that for make real money you have to use CatBoost

[Javier Abellán @javier_a2]

@manu__martinm En 10 habrá pinchado la burbuja y se descubrirá el talento de verdad del sector tech 😁 (tanto mayores de 25 como menores) https://t.co/2OCwsQQoOo

[kanav @kanavtwt]

Here Comes Another Bubble (AI Edition)

[kanav @kanavtwt]

Here Comes Another Bubble (AI Edition)

[Javier Abellán @javier_a2]

I'm the only one who is pissed off at @huggingface for taking the https://t.co/Uf2LzgzGlG page down? I used to use it a lot to check SOTA models 😠😠😠

[Javier Abellán @javier_a2]

@boardinggate @javilop Más bien al revés, los usuarios de Starlink ya se reiron de nosotros el 28 de abril de 2025 😁

[Javier Abellán @javier_a2]

@MurciaDev Muchas gracias por la oportunidad, espero que os guste y no aburriros mucho con la chapa que os voy a contar 😁

[Javier Abellán @javier_a2]

@chrisalbon Theorically the training error is zero at interpolation threshold because the model is "big enough" to completely learn the training set. In practice is probably near zero like you plotted.

[Javier Abellán @javier_a2]

@pablomarc0s @flopezluis @grok @Fluendo @rfleury @vkrajacic Yo creo que al igual que la gente sin saber de programación puede hacer cosas muy chulas en Python, o C. Nosotros (programadores) podemos hacer lo mismo con C porque ya no da pereza, (donde en la era pre IA la opción rápida era solo Python). Por lo tanto espero un auge de C/C++