vulture916

Granite is super good small parameter lineup for local tasks.

This is an awesome idea and implementation, thanks!

thanks for this - it's like the awesome plannotator... on a bit of steroids and, at least to my brain, cleaner.

The article about (another) the end of the world is behind a paywall, the tweet about it is behind a login screen - guess I'll find out when it happens!

In my world, that'd be illegal.

Jev = $0.042/m input, output free Clef = $0.24/m input, no output price listed

At 300 tokens per call, you'd get:

One million decisions on Jev cost about $12.60. One million decisions on Clef cost about $72.

Would probably make sense to self-host Clef, if you have the capability/resources. If not...

Why this one vs. 400 other registries?

Just sent this to my CEO - thanks!

Even though all these "ADE"s are mostly the exact same (Superset, T3, Conductor, Paseo,random Reddit-dude,etc.). I'm liking Orca largely because of the built in browser with ability to select an object(s), write a comment and send it back to the agent. I had built my own solution, but Orca makes it even easier and convenient.

Either I'm an idiot or am missing something entirely (maybe a demo video would help?)

I've got to send it to your server so you can spit it back when I say a fancy word? What's worse, I pay you to just store/retrieve text? Why wouldn't I just write to a file?

Check out Tailscale - they have TUNS + share the source files with someone else in tailnet.

I haven't done much long-run OCR, so unsure of the current state, but it would seem they overcome this (from their paper):

"A widely held view is that employing a large language model (LLM) as the decoder allows the model to leverage the prior distribution of language, leading to improved OCR performance. However, the downside is equally evident: as the output sequence lengthens, the accumulated KV cache drives up memory consumption and progressively slows down generation."

It's gotten really good, just slow as all hell.