Latest Search
Quote
| Back Zoom + Zoom - | ||||
|
DeepSeek Releases Open Source Multimodal AI Models; Text-to-picture Tests Reportedly Outshine OpenAI
Recommend 46 Positive 87 Negative 23 |
||||
|
||||
|
DeepSeek has released a series of open source multimodal AI models called Janus-Pro and JanusFlow respectively, Chinese media cited an AI community, Hugging Face. DeepSeek's presentation displayed that Janus-Pro-7B has improved its accuracy over its predecessor, Janus, with 80% and 84.2% in the benchmark text-to-picture tests of GenEval and DPG-Bench, respectively, which were higher than OpenAI's DALL-E 3 and Stable Diffusion models. Janus has an accuracy rate of 61% and 79.7%. AAStocks Financial News |
||||
Copied to Clipboard
 |
||||
