New #AI tool #MiniGPT4 is an open-sourced model performing complex vision-language tasks like GPT-4.
Upload an image, the AI analyzes it, and you can chat with the AI to ask for descriptions, pinpoint problems, code a website from a handwritten note, identify people, design ads, write a poem, and more...
➡️ https://minigpt-4.github.io https://nostr.build/av/nostr.build_f16effaf1205bbf7d61ca82ec7a137471d9aa5c4dfd7f186dcc216140b88800c.mov
Yes nice. Have you managed to build anything with auto gpt? I always get stuck in a loop with no real outcome
Heard that’s pretty common. I’ve been following the development of Auto-GPT forks and plugins designed to work with local models like Vicuna and GPT4All (so you don’t have to send API requests and money to OpenAI). So far they’re running into major output quality issues, but once it’s figured out the permissionless innovation should be huge!
Yeah possibility is insane, I have not heard of Vicuña, will have a look 🫡
