Post by G0LDSHARK in Silverpine comments

Viewing post in Silverpine comments

GLM-4 seems to give up when i try to talk to npc's just ... Forever it seems. then a lmm error or something happens

Does your GPU have 24+ gb of vram?
There also seems to be an unfixed backend issue with Vulkan/AMD GPUs for this specific model. I believe the workaround they currently use might result in context processing so slow that the API call times out, but I have no way of testing or fixing this.

G0LDSHARK31 days ago

nvida and i got 32 gb so yes

Three Eyes Software31 days ago

Does this happen immediately or after several inputs? Does it print anything special on the KoboldCPP command prompt?

G0LDSHARK31 days ago

Immediate as it fails if i talk to any npc. instantly. and no for the second part.

Three Eyes Software31 days ago

Do you also have this issue with Nemo?

G0LDSHARK31 days ago

Nope only GLM-4 has the problem the others are just fine

Three Eyes Software31 days ago

I will investigate this.

Three Eyes Software30 days ago(+1)

After testing the model on cloud based 3090s/5090s, I've come to the conclusion that the upstream GPU implementation for this model is completely broken.
This didn't occur to me during testing as I don't own a GPU with 24 gb of vram myself, and as such did all the testing using slow CPU inference only.

I will look for an alternative 24 gb model, and add GLM-4 back in once it's properly implemented upstream.

G0LDSHARK30 days ago(+1)

on a side note any chance we can get a way to make items? The ai already seems to have the ability to. but a way to make custom items, armor, and weapons would be nice.
thanks for listening to your players!

Three Eyes Software29 days ago

I will experiment with it.

View more in thread

itch.io

Viewing post in Silverpine comments