We just shipped an update to openevals, our open-source library for evaluating LLM applications. Multimodal support: pass images, audio, and PDFs
MULTIMODAL AI
-
Visual Imitation Learning: Training AI Agents on Expert Videos
By
–
Visual imitation learning: Guidde trains AI agents on human 'expert video' instead of documentation https://
buff.ly/r2014RE
#AI #MachineLearning #DeepLearning #LLMs #DataScience -
LatentMorph: AI Image Generators Learn Reasoning Like Humans
By
–
What if AI image generators could truly 'reason' and adapt their creative process, just like humans? Researchers from HKUST, HKUST(GZ), NWPU, and Knowin just unveiled a groundbreaking approach. Their new method, LatentMorph, introduces a novel way for Text-to-Image models to
-
Mastering PyTorch: Create and Deploy Deep Learning Models
By
–
Mastering PyTorch: Create and deploy deep learning models from CNNs to multimodal models, LLMs, and beyond – http://amzn.to/40IFEQR via @PacktDataML #AI #ML #MachineLearning #DataScience #DataScientist #GenAI
-
New version of MuleRun AI agent tool offers enhanced workflow automation
By
–
Long time fan of @MuleRun_AI, but V2's in a league of its own.
— Charly Wargnier (@DataChaz) 16 mars 2026
Imagine a secure, supercharged OpenClaw 24/7 on a powerhouse cloud server.
It:
→ plugs into "180,000+ agent skills 🤯
→ learns all your daily workflows
→ gets smarter over time
I've added cool links in the 🧵↓ https://t.co/DDDsT4F0YvLong time fan of @MuleRun_AI
, but V2's in a league of its own. Imagine a secure, supercharged OpenClaw 24/7 on a powerhouse cloud server. It:
→ plugs into "180,000+ agent skills → learns all your daily workflows
→ gets smarter over time I've added cool links in the ↓ -
Pokémon Go’s 30 Billion Images Secretly Built Massive AI Dataset
By
–
143 millions de personnes pensaient chasser des Pokémon.
— VISION IA (@vision_ia) 16 mars 2026
En réalité, elles construisaient l'un des plus grands datasets visuels du monde réel de l'histoire de l'IA.
Niantic vient de révéler que les photos et scans AR collectés via Pokémon Go représentent plus de 30 milliards… https://t.co/lEauNwHvpE143 million people thought they were hunting Pokémon.
In reality, they were building one of the largest real-world visual datasets in the history of AI.
Niantic just revealed that the photos and AR scans collected via Pokémon Go represent more than 30 billion images.
And these -
Google adding Google Maps grounding to Gemini API
By
–
Google is working on a new "Grounding with Google Maps" tool for AI Studio. This option will let Gemini APIs easily access "Location results from Google Maps". Maps coding
-
3D scanning technology creates dreamlike replay experience
By
–
When your 3d scan is like replaying a dream https://t.co/enPwHdFn48
— Bilawal Sidhu (@bilawalsidhu) 15 mars 2026When your 3d scan is like replaying a dream
-
AI Assistants Map Inner World Through Spatial Intelligence
By
–
AI assistants are mapping our inner world.
— Bilawal Sidhu (@bilawalsidhu) 15 mars 2026
Spatial intelligence is mapping the outer one.
Layer them together and this is what you see:
pic.twitter.com/opdGA6fYmwAI assistants are mapping our inner world. Spatial intelligence is mapping the outer one. Layer them together and this is what you see:
-
Tesla Optimus Humanoid Robot Dancing with Elon Musk
By
–
Elon Musk and TESLA Optimus Dancing l?
— Amitav Bhattacharjee (@bamitav) 15 mars 2026
pic.twitter.com/YVjGe7o1fD#tesla #optimus #humanoidtech #humanoid #robot #Robotics #AI #TechRevolution #TechInnovation #ArtificialInteligence #PhysicalAI@AlbertoEMachado @Eli_Krumova @postoff25 @Khulood_Almani @anand_narang @NutritiousMind…Elon Musk and TESLA Optimus Dancing l? #tesla #optimus #humanoidtech #humanoid #robot #Robotics #AI #TechRevolution #TechInnovation #ArtificialInteligence #PhysicalAI @AlbertoEMachado @Eli_Krumova @postoff25 @Khulood_Almani @anand_narang @NutritiousMind