Alex "mcmonkey" Goodwin
Alex "mcmonkey" Goodwin
I've been intending to figure out getting this working in the webui, but, uh, the limitation is I don't have a multi-GPU setup to test with currently.
That error is a code error within the underlying peft/Transformers trainer itself that happens if you interrupt after it starts training but before it completes a single step. I tried...
... I'm at a loss actually, something is indeed wrong with saving. `adapter_model.bin` is outputting a 443 byte file - an empty pickle with no data. Not even just when...
No I didn't replicate any VRAM issues while testing. EDIT: External users (on Discord) are saying they get the VRAM spike on peft==0.2.0 actually. Yes I'm not aware of any...
Oo, that explains why I never saw it! I haven't tried updating bitsandbytes for a while. That's good info, thank you, we can notify the other users getting VRAM spikes...
See my answer explaining this @ https://github.com/oobabooga/text-generation-webui/discussions/839#discussioncomment-5557486
That error message indicates you don't have GPTQ installed. See for info. It likely won't work for anyone not using an nvidia GPU right now. CPU models might be a...
@UrielCh llama-cpp-python is for CPU running, this error message comes from GPTQ which is for GPU running. You're likely trying to load a GPU model by mistake instead of a...
There's a lot of CPU-compatible models out there, there's a download list for popular CPU models @ https://rentry.org/nur779 (disclaimer: I have no idea who maintains that) You can recognize that...
Yes, yes that does, that's a CPU model.