MrHćtr retweeted
Here is a good blog post: unstract.com/blog/understand…
MrHćtr retweeted
Kind reminder: Small LMs aren’t dead. Old BERT, RoBERTa, T5, etc. can still be efficient (and enough) for your use case.
MrHćtr retweeted
Thank you @karpathy for the shout-out to llms.txt in your wonderful talk about software 3.0
youtu.be/LCEmiRjPEtQ?si=NEu9…
MrHćtr retweeted
🇪🇸Durante los últimos dos años, el equipo de @PLD_Space ha estado desarrollando electrodeposición de cobre y níquel, soldadura de Inconel, inspección de rayos X y nuevas tecnologías de fabricación END para cumplir con los requisitos de diseño de propulsión de MIURA 5. Después de la fabricación de varios conjuntos de cámaras de empuje a escala completa y subescala durante este tiempo de desarrollo, la producción completa del motor ahora está en marcha; 4 motores de la familia #TEPREL-C ahora están en fabricación para el comienzo de la calificación del motor a fines de junio.
🇬🇧Over the past two years the @PLD_Space team has been developing on Copper and Nickel electroplating, Inconel welding, X-ray Inspection and novel NDI manufacturing technologies to meet the propulsion design requirements of MIURA 5. Following the manufacture of several sub- and full-scale thrust chamber assemblies over this development time, full engine production is now in operation; 4 #TEPREL-C family engines are now in manufacture in preparation for the beginning of engine qualification in late June.
MrHćtr retweeted
Reminder that it is possible to demonstrate the regular manipulation of 'election' results in Russia based on the official data
MrHćtr retweeted
Telefónica y Vithas prueban el blindaje frente a ataques cuánticos en dos hospitales telecomkh.info/?p=20314 @telefonica @Vithas @LuxQuanta #QoolNet #BLTBiolight #ACL #ComputaciónCuántica #AtaquesCuánticos #ciberseguridad #MWC #MWC25
MrHćtr retweeted
We've released a new multilingual, open-source visual embedding model and training set on Huggingface!
vdr-2b-multi-v1 produces single dense vectors for visual document retrieval, enabling efficient large-scale systems. Key features:
➡️ Trained on 5 languages (IT, ES, EN, FR, DE)
➡️ 70% fewer tokens = 3x faster inference for the english-only model
➡️ Matryoshka Representation Learning for flexible dimension reduction
➡️ Full training set is open-source
Learn more:
📝 Blog post: huggingface.co/blog/vdr-2b-m…
🤗 Model + LlamaIndex integration: huggingface.co/llamaindex/vd…
MrHćtr retweeted
Más importante que minar y construir baterías es tener el conocimiento del proceso.
#NoALaMinaCC
caixinglobal.com/2025-01-03/…
MrHćtr retweeted
Está todo relacionado
MrHćtr retweeted
Maya, an open-source Multilingual Multimodal model by @CohereForAI
• Multilingual dataset
• Toxicity-free version
• 8B parameter model
• Open dataset 🎊
• Apache 2.0! 🎉
MrHćtr retweeted
correr con el coche ahorra menos tiempo que la esperanza media de vida que se pierde 😅
MrHćtr retweeted
Most companies are still experimenting with AI.
We are still far, far away from finding product-market-fit for all of the recent AI progress we've made.
Talk to people outside your online bubble:
99% of AI applications still use boosted trees and convolutional neural networks.
