darkeox

joined 1 year ago
[–] [email protected] 1 points 1 year ago

My bad. I think I confused this with the previous popular Unigine benchmarks.

[–] [email protected] 3 points 1 year ago (2 children)

For relative comparison sure but isn't it starting to become irrelevant these days? No Vulkan nor Raytracing nor Compute.

[–] [email protected] 7 points 1 year ago* (last edited 1 year ago) (1 children)

Best shot:

There was a self contained AppImage around once but it had bugs. Maybe finding it, upgrade the Wine/DXVK/VKD3D inside and try...

EDIT: Last known best compatible version is 2021. And from my recalling, it works best if installed on Windows and then moved on Linux (plus all the littered files it creates around Windows).

[–] [email protected] 4 points 1 year ago

That's because they imply DLSS 3.5 usage. Which isn't too unfair but a bit misleading.

[–] [email protected] 2 points 1 year ago

Alright, thanks for the info & additional pointers.

[–] [email protected] 0 points 1 year ago

Well the thing with those "enabled EAC on Linux to see where it gets us" is it's non-binding and non-commital. And it's made explicitely that way so that support cannot be demanded from Linux users unlike Windows users who are explicitely mentioned in the systems supported by the game.

We legally don't have any ground to be supported the same as Windows users.

[–] [email protected] 0 points 1 year ago (2 children)

That's the problem with non official support. You're basically in an unending beta-testing phase. Theres' no easy solution here I'm afraid.

[–] [email protected] 2 points 1 year ago (2 children)

Ah thank you for the trove of information. What would be the best general knowledge model according to you?

[–] [email protected] 1 points 1 year ago

This. It's not easy or trivial but as a long term strategy, they should already plan investing efforts into consolidating something like Godot or another FOSS engine. They should play like you calm down an abuser you can't just escape yet while planning their demise when the time has come.

[–] [email protected] 2 points 1 year ago (4 children)

Don't be sorry, you're being so helpful, thank you a lot.

I finally replicated your config:

localhost/koboldcpp:v1.43 --port 80 --threads 4 --contextsize 8192 --useclblas 0 0 --smartcontext --ropeconfig 1.0 32000 --stream "/app/models/mythomax-l2-kimiko-v2-13b.Q5_K_M.gguf"

And had satisfying results! The performance of LLaMA2 really is nice to have here as well.

[–] [email protected] 2 points 1 year ago (6 children)

Thanks a lot for your input. It's a lot to stomach but very descriptive which is what I need.

I run this Koboldcpp in a container.

What I ended up doing and which was semi-working is:

  • --model "/app/models/mythomax-l2-13b.ggmlv3.q5_0.bin" --port 80 --stream --unbantokens --threads 8 --contextsize 4096 --useclblas 0 0

In the Kobboldcpp UI, I set max response token to 512 and switched to an Instruction/response model and kept prompting with "continue the writing", with the MythoMax model.

But I'll be re-checking your way of doing it because the SuperCOT model seemed less streamlined and more qualitative in its story writing.

 

Hi,

Just like the title says:

I'm try to run:

With:

  • koboldcpp:v1.43 using HIPBLAS on a 7900XTX / Arch Linux

Running :

--stream --unbantokens --threads 8 --usecublas normal

I get very limited output with lots of repetition.

Illustrattion

I mostly didn't touch the default settings:

Settings

Does anyone know how I can make things run better?

EDIT: Sorry for multiple posts, Fediverse bugged out.

 

Des trains pleins longtemps à l'avance et ils n'en rajoutent pas ? Soit il aiment perdre de l'argent soit ils font ça pour nous pourrir la vie mais dans les ...

view more: next ›