I like this stuff.
https://video.nostr.build/84ac16847bc06c56cb5441cebcf7c871b940503e074df7b3cde40918762529ce.mp4
Discussion
Love it! It's publicly available?
Just added it to the bot. It's the new open source stable video diffusion model (still a bit in test phase)
Testing it! ⚡
Please note it requires an image as input.
I see, my first test was empy and it fails, maybe can be good to catch that before take the payment, but in any case love it!
https://video.nostr.build/3f4b0a0272897980a3dbb19fa1ed51334c24e56bcce3ab95cff49dab3bb29604.mp4
Yes on my list! But I can only do this for DVMs I control. I zap you the fail attempt back.
No worries man! Take it like a tip 🫂💜
Btw you can also chain this with the media converter and select
8 https:// linktovideo -format image/gif to create a gif out of it
Thats so cool 👏
Are you using your framework for dvms that I saw on GitHub?
Yes. I'm currently rebuilding the whole thing to be a framework (already kind of is by now) so that people can hopefully use and extent DVMs to their own interest/models/services easily. I'm providing some examples there on how to contact different apis and how to run stuff locally. I want it to be a bridge between nostr and whatever service we want it to connect to. :)
That's beautiful man! I'm just starting to dig deeper into nlp, I see that at the moment there is only support for openAI and replicate.com, it would be really cool to integrate the ollama api for example, so anyone could use their own infrastructure to run a dvm. It could also be interesting to integrate langchain, for example in the case of image-to-video to gif. In any case I hope to have some more free time to see if I can do some pr or play a bit more with the framework.
There's also a branch with local models like whisper, SDXL, image upscaling, image analysis, but need to finish that up and write some documentation for it to make it useable for other people.
But yes, adding more stuff is on the list, besides extending the current capabilities of DVMs in general. And of course happy if people want to contribute or even simply run DVMs 🤙
First version of ollama API added :)
Wow, that was fast! Ill try to run it tomorrow, there is a branch in the repo?
It's in main. I'm using the llmlite lib as it also supports other backends which comes in handy. You find it in tasks/textgeneration_llmlite.py. You can also run the task files alone without the bot/other DVMs by just running it from the main folder. python tasks/nameofscript.py. let me know if I can assist with anything. Still in the process of making everything smooth for people to use.
Beautiful
First post I see in my feed gets the sats today this was it