b24f0 - 2y
llava can generate tags for any pics you post some examples here nostr:npub14ewxuzm5vcxjr9xgy494ctyz2em9w6lq99ehn73q2gc3ff4gt6rs3nhz9g here is how install oobabooga webui install llava-7b-v0-4bit-128g start textgen webui with multimodal python server.py --api --chat --listen --extension multimodal --model llava-7b-v0-4bit-128g go to chat and send your pic, it will comment that add message "write hashtags for this picture" format output latest bk.7z has bash tools for this https://bafybeibg3k6qqa5gbphp4gih67ojlpcsm5ugstcwsnk7yhoenhz4u3e2hm.ipfs.dweb.link/# eg 1) upload next picture from upload/rozenmaiden_suigintou/ 2) generate tags with llava in addition to those specified manually 3) post it to nostr user account dev, defined in file config.local 1, 2, 3 all with one command: ./upone.sh dev "rozenmaiden_suigintou" "#rozenmaiden #suigintou #hardess #girl #cute #sexy #art #beautiful" ./upone.sh <user> <directory> <tags> (additional ai tags will be generated based on image file) // there is also new llama2 based llava model, but so far no success getting it to work in ooba webui // also blib models might be more accurate at reading images, but tags can be little creative ofc
