Javier

AI/NN developer. Always learning new things.

[Javier @javi_22_dev]

Just released a weekend project at https://t.co/8mhdmvXRhG, would appreciate feedback!

[Javier @javi_22_dev]

@sflorimm Logs

[Javier @javi_22_dev]

@0xSero Don't think so

[Javier @javi_22_dev]

@kimmonismus Looks like a fine-tune, it's fine, but not a big deal

[Lisan al Gaib @scaling01]

Prediction time:

- Europe will absolutely learn nothing from this incident. At best they are going to invest an additional 2B Euros in some robotics or image gen startup, because they think that's somehow relevant

- China will get even more AGI pilled and restructure their entire economy for one purpose

[David Cramer @zeeg]

codex writes the most digusting code

idk who's responsible for pre-training over there but you gotta flip the script

[Zephyr @zephyr_z9]

"OpenAI will leapfrog Anthropic with their 15-20T model"
🤣🤣🤣🤣
OpenAI's pretraining team is a complete shitshow
All the good pretraining people are at Anthropic

[Javier @javi_22_dev]

@Nature I'm not. Still running Opus 4.8.

[Javier @javi_22_dev]

@marlytokenman @scaling01 They wouldn't know that we knew. Now they do

[Javier @javi_22_dev]

@mao888ahw @GoogleDeepMind @googlegemma I did, they've been working on it for some time. It's a fascinating paradigm.

[Javier @javi_22_dev]

@Cryptoalterno @nicobustelo__ Desde el 3.5 Sonnet no hay modelo de OpenAI que me parezca mejor que los de Anthropic.

[Javier @javi_22_dev]

@scaling01 The test time slope has scaled up. Big model smell.

[Javier @javi_22_dev]

@kimmonismus

[Javier @javi_22_dev]

@mavihsk @kimmonismus Yeah, same here

[Javier @javi_22_dev]

@NVIDIAGeForce #RTXPowersPlay I'd use it to train so many NNs

[Javier @javi_22_dev]

@LegalMalvado De qué va el proyecto? Si no es información sensible, tira el modelo de embedding y la base de datos en local, y usa un modelo potente por API

[Javier @javi_22_dev]

@VraserX You think the power of a phone / laptop is worth the latency? Maybe it'll get better as network speeds increase, but still...

[Javier @javi_22_dev]

@ideogram_ai Why using a VLM and not a text-only LLM for the text encoder? Just curious

[Javier @javi_22_dev]

@Rishii25 @bindureddy Not frontier level

[Javier @javi_22_dev]

@OfficialLoganK I could use some GPU hours tbh