Cool thanks for sharing. I would've assumed you would benefit from running llama.cpp to better utilize your available cpu now that you're running dedicated vRAM. My understanding might be wrong, but I think you might have some options on your hand while running those GPUs + CPU. For me, on unifi...

Good luck. And be fast. Prices go up

Do it!

Get em while there hot! Fresh off the line! :wisp_lazor:

I totally read guns at first glance....

With https://deno.com/deploy. The event is signed locally with created_at = publish time and handed off to a Deno Deploy hold-and-forward service, which persists it in Deno's key value database until due. A cron job runs every 60 seconds, picks up due events, and publishes them via WebSocket to the ...

16gb of VRAM for about $160... Yes.

I do like how Amethyst shows chat room responses from your follows in the conversation tab because then you see a convo is happening and also serves like billboard for the stream in your feed.

Can it have a bridge to Openclaw?

Did you Ralph it?

https://blosstr.com/6522efb5ab395657a700f49d290f1f1f199572c1714a8cc5cf8a3df043b1c49f.gif

https://media1.tenor.com/m/Cehn_T-iwDMAAAAC/oh-my-god-okay.gif

I like qwen

Whatcha running?

Every time I try, it won't let me https://npub1ehkvx8rdjsrwnf7kkqr8gy42vcwe6vwgqdwrl4juqccp689v8wfqr3mz4s.blossom.band/1811678416281a7033b06c223c6b83ba2ec75969794ef7043bf38ea4f5113c85.jpg

Clanked 😁

I’ll totally agree to this!

But good luck training up and running inference on a model that isn't a moron...? This is my skepticism with AI, it seems to be an inherently centralizing technology...

Microsoft accepts your challenge!

True, but hardware is an obvious attack vector. Access is already limited right now due to economic reasons. So theoretically, you could be at some point. Whether by ban, regulation, or market forces. Especially as computing is pushed to the cloud. I can see that accelerating with hard price increas...

ttrue dat

i can't let you do that dave

ABSOLUTELY!!

Exactly but training stuff may or not may require powerful computing

similarly mesh llm and routstr type networks seem important

Will an 'AMD Strix/Krackan/Strix Halo Neural Processing Unit' do? :P Here's searching for smol things to run on it.

Yet.