Deepseek just dropped an Any-to-Any, MIT licensed, multimodal model (both images and text) 👏 Outperforms DallE 2/ SDXL in generation and Llava 1.5 7B in multimodality 🔥 **Results** 📊: • Outperforms unified models on multimodal understanding benchmarks: - MMBench: 69.4 (vs… https://t.co/0jEz4n3zTM https://t.co/GO7tuP9pCG
— Rohan Paul (@rohanpaul_ai) Oct 18, 2024
from Twitter https://twitter.com/rohanpaul_ai
October 18, 2024 at 01:03PM
via IFTTT
No comments:
Post a Comment