A naukri.com initiative
TechCrunch
1M
186
Image Credit: TechCrunch
AI2’s Molmo shows open source can meet, and beat, closed multimodal models
- AI2 released Molmo, a visual understanding engine, the multimodal AI model aims to match Google's best models while being open-source and free.
- Molmo can identify and answer questions ranging from everyday situations to object identification.
- Molmo performs on par with the likes of GPT-4o, Gemini 1.5 Pro and Claude-3.5 Sonnet, while being much smaller; at a tenth of their size, it approaches their level of capability with a model that’s a tenth of their models.
- AI2 curated and annotated a set of 600,000 images only, compared to billions used in other models.
- Molmo produces image descriptions that are conversational and useful, an interesting annotation method is used where people describe the images out loud.
- Molmo's dataset and code are completely free and open-source, empowering developers to make AI-powered apps, services and experiences without seeking permission or paying tech giants.
- AI giant players are themselves lowering their prices, raising hundreds of millions to cover the cost, while smaller players like AI2 successfully release free open-source models like Molmo.
Read Full Article
11 Likes
For uninterrupted reading, download the app