Ollama Now Officially Supports Llama 3.2 Vision - Talk with Images Locally

[@fahdmirza]

🔥How to Upgrade Ollama Version in Linux Windows and Mac - ruclips.net/video/338CNrJgN08/видео.htmlsi=a8FWgEqzPOPQ8pBn

[@xw33b36]

Woo! lets goo! I forked a copy of open-webui to use clip for all image uploads so any model could parse images but this will be so much easier. just need bigger vision models now! 11b still so tiny. Also, Fahd, you are truly my favorite machine learning youtuber. Can you do us a video soon with 5 best LLM and multi-modal LLM's right now?

[@xw33b36]

Also my friend, have you done a video on deepspeed? It would be interesting to see how people could make use of it to create cuda nodes on local network from multiple machines.

[@fahdmirza]

Sounds great! and sure, I will look into best LLM video, though it changes so rapidly the shelf life would be so little. Also thank you for the Super Thanks. You just made my day.

[@Mnbir]

Thanks

[@fahdmirza]

Thanks a ton for Super Thanks. Much appreciated 🙏

[@andrepaes3908]

What's the maximum resolution of pics you can upload in the model?

[@fahdmirza]

1120x1120 pixels

[@ajarivas72]

​@@fahdmirza your videos are awesome

[@andrepaes3908]

@@fahdmirza Wow that's pretty good! thanks for the reply!

[@enricollen]

Thanks ❤

[@fahdmirza]

You're welcome 😊

[@bossgd100]

The LLM answers are fast, what GPU are you using ?

[@fahdmirza]

I mentioned that in the video. thanks

[@bossgd100]

@fahdmirza I will watch it again more carefuly thank you. I am not native english speaker

[@fahdmirza]

@@bossgd100 thank you, let me know if you cannot find it, I will fish it out for you.

[@alejandrof9816]

Thanks for the video! I've tried Llama3.2 vision on AnythinLLM but even though the chat works, the vision doesn't seem to recognize any images. Do I need any extra configuration?

[@fahdmirza]

yes you need to fine-tune on a pertinent dataset

[@alejandrof9816]

@fahdmirza I'll keep that in mind, thanks for the reply!

[@bossgd100]

Can you create a bot from it ? like a game bot that click and type in a web browser game ?

[@fahdmirza]

yes

[@alisaid3745]

Its cutoff date is December 2021 means 3 years.
Looking a module its cutoff date on 2024

[@roshanYadav-y8f]

I have nvidea Rtx 3050 can i use

[@kumarmanchoju1129]

yess

[@saadirfan2358]

When will ollama support tti ?

[@fahdmirza]

thats the question for ollama team :)

[@fineCoder.]

can I run this in 1650 (16 gigs)

[@fahdmirza]

yes

[@ShaunyTravels.]

Ok thats great but what about a image containing text?

[@fahdmirza]

I have similar videos on the channel, please search, thanks

[@viangelo4z595]

my gpu is dead

[@fahdmirza]

how much vram you have

[@viangelo4z595]

@fahdmirza The total ram of my computer is 32 🥲

[@fineCoder.]

can i run this in 16 gigs 1650