sayzard1d ago

Sam Hogan (@samhogan)

NVIDIA가 'frontier models'를 공개한다고 알리며 오픈소스 경쟁력을 지키려는 의지를 보였습니다. 아직 폐쇄형 모델만큼 성능이 되진 않지만, 좋은 베이스 모델에 경량 후처리만으로도 많은 과제에서 SoTA 수준 또는 그 이상을 달성할 수 있다는 전망을 제시합니다. 오픈소스 모델 생태계의 성능 향상과 경쟁 촉진을 시사합니다.

https://x.com/samhogan/status/2033350986056831154

#nvidia #opensource #foundationmodels #llm

Sam Hogan 🇺🇸 (@samhogan) on X

NVIDIA is releasing frontier models & has 4.38 trillion reasons to make sure open source stays competitive They’re not quite as good as the closed source stuff but I’m confident they’ll get there Good base model + some light post training get you ~SoTA or beyond on many tasks

X (formerly Twitter)
Ricardo Dantas1d ago

Today I'm shipping BragLog! 🎉

I built this app to solve a problem I kept hitting: performance review time comes and I can't remember what I did two months ago.

If you've ever frozen at "what did you accomplish this quarter?" — this is for you.

https://braglog.app

#iOSDev #SwiftUI #Swift #FoundationModels #IndieDev #Apple #Launch

Show threadsbm  3d ago

- TranslationSession.Strategy.highFidelity: https://developer.apple.com/documentation/translation/translationsession/strategy/highfidelity
- TranslationSession.Strategy.lowLatency: https://developer.apple.com/documentation/translation/translationsession/strategy/lowlatency

#iosdev #SwiftLang #SwiftIUI #FoundationModels (3/3)

highFidelity | Apple Developer Documentation

A translation strategy that provides more fluent translations using Apple Intelligence.

Apple Developer Documentation
sayzardMar 9

Ilir Aliu (@IlirAliu_)

파운데이션 모델이 점점 '더 똑똑해지고' 있지만 업계는 여전히 '데모 시대'에 머물러 있고, 매끄러운 영상 데모가 현실 작동을 보장하지 않는다는 지적입니다. 특히 로봇 분야에서 'Validation Gap'(검증 격차)을 어떻게 해결할지에 대해 긴 논의를 예고하는 내용입니다.

https://x.com/IlirAliu_/status/2030717056426057781

#foundationmodels #validation #robotics #ai

Ilir Aliu (@IlirAliu_) on X

Foundation models are getting "smarter," but we’re still stuck in the "demo era." A smooth video doesn't mean a robot is ready for the real world. How is the industry tackling this "Validation Gap"? 🧵 This is gonna be a long one ❗️❗️ [Save this so you can find it when you

X (formerly Twitter)
sayzardMar 4

merve (@mervenoyann)

Qwen이 최고의 베이스 모델을 제공해 여러 연구실과 오픈소스 AI 생태계에 큰 변화를 일으키고 있다는 내용입니다. 많은 팀들이 Qwen을 기반으로 모델을 구축하며 오픈소스 AI 전반에 충격적인(재편을 촉발하는) 변화가 발생하고 있다는 관찰을 공유하고 있습니다.

https://x.com/mervenoyann/status/2028924000517177580

#qwen #opensource #foundationmodels #ai

merve (@mervenoyann) on X

just yesterday I posted this Qwen delivers the absolute best base models that everyone builds on them quite shaking changes for every single lab in open-source AI tbf

X (formerly Twitter)
sayzardMar 3

Min Choi (@minchoi)

DeepSeek V4가 이번 주 공개된다는 발표. 1조(트릴리언) 파라미터 규모의 멀티모달 모델로 텍스트·이미지·비디오를 지원하며 Huawei 및 Cambricon 칩에 최적화되어 있음. 오픈소스이며 중국에서 무료로 제공된다고 명시되어 대규모 공개 모델로서 중요함.

https://x.com/minchoi/status/2028811096602120310

#deepseek #multimodal #opensource #foundationmodels #huawei

Min Choi (@minchoi) on X

🚨DeepSeek V4 drops this week. Trillion parameters. Multimodal. Text + image + video. Optimized for Huawei/Cambricon chips Open source. From China. For free.

X (formerly Twitter)
Bob CarverFeb 27
The Model That Knows Too Much: How AI Can Leak What It Learned
https://youtu.be/pwA5nASJpoo #Cybersecurity #ArtificialIntelligence #AIsecurity #ModelInversionAttacks #DataPrivacy #MachineLearning #AISafety #Infosec #DigitalRisk #FoundationModels #DataProtection #CyberRisk
sayzardFeb 25

Lukas Ziegler (@lukas_m_ziegler)

Physical Intelligence(@physical_int) 기반의 로봇들이 실제 세탁소에서 세탁물 접기 작업을 수행 중이라는 알림입니다. 해당 기업은 배포 파트너들과 협업해 자사 파운데이션 모델(또는 '로봇 브레인')이 다양한 실세계 환경에서 성능을 발휘함을 시연하려 하고 있습니다.

https://x.com/lukas_m_ziegler/status/2026603886866256087

#robotics #foundationmodels #physicalintelligence #deployment

Lukas Ziegler (@lukas_m_ziegler) on X

Robot brain that makes robots more capable! 🧠 @physical_int -enabled robots fold laundry in production at real laundromats. Physical Intelligence is collaborating with deployment partners to demonstrate their foundation models perform in real-world settings across different

X (formerly Twitter)
Hylke RozemaFeb 25
Five Architectures for Time Series Forecasting with Large Language Models
Large Language Models are increasingly being applied to time series forecasting. Not as chatbots, but as prediction engines that leverage the pattern recognition capabilitie
https://www.hylkerozema.nl/2026/02/25/five-architectures-for-time-series-forecasting-with-large-language-models/
#DataScience #MachineLearningEngineering #DataScience #Forecasting #FoundationModels #LLM #MachineLearning #TimeSeries #Transformer
Five Architectures for Time Series Forecasting with Large Language Models – Hylke Rozema

Harald KlinkeFeb 23
Europe’s competitiveness will depend on linking its foundation models with industrial applicability, governance standards, and sectoral integration. SOOFI: ~100B parameter, reasoning model and agentic applications. #SovereignAI #EuropeanAI #FoundationModels

SOOFI Launches: Europe’s Path ...
SOOFI Launches: Europe’s Path Towards Its Own AI Language Models - L3S Research Center

With SOOFI (Sovereign Open Source Foundation Models), one of the most ambitious initiatives to strengthen European AI sovereignty is now underway. The project aims to develop an open AI language model with around 100 billion parameters that meets European values and regulatory requirements. Six leading research institutions are involved – including the L3S Research Center.

L3S Research Center - Trustworthy AI & Digital Transformation