#models load error

1 messages · Page 1 of 1 (latest)

woeful igloo
#

how can fix this error

desert stream
#

it looks like you're loading a (standard/unquantized) pytorch model, these should be in a subfolder in models, and then use the Transformers loader.

But it's likely that you don't want to load such a model at all unless you have tons of (V)RAM

woeful igloo
desert stream
#

You should start by looking at what kind of system you have (GPU, VRAM, RAM, etc) and then decide on what you want to do, so you can actually choose a suitable model.

woeful igloo
#

GPU: RTX 4080 12GB, CPU: i9-13980, RAM: 32GB. Are there any recommended models?

desert stream
#

Probably start with a 13B GPTQ model

desert stream
#

Select ExLlamav2 as the Model Loader (or ExLlamav2_HF, I think that's the recommended one think_blob)

dawn needle
desert stream
#

thonk Trying to help someone setting up their webui for the first time, it needs to just work and not need anything fancy

woeful igloo
#

I downloaded the model through Hugging Face and placed it in the 'models' folder. I've tried all the model loaders, but I still encounter errors after loading. Does it only work properly through downloading the model or using LoRA?