skip downloading model weights

[sieste]

I have already downloaded the 7B model weights. How can I avoid redownloading them with npx dalai llama?

[marcuswestin]

Fixed in #16

[ekp1k80]

You have these 2 options
#16
#5

I recomment the 16. I download the repo and run yarn just:run You have to move the models folder (the model you already download) to llama.cpp

I did'nt be able to run it so. The file convert-pth-to-ggml.py is crashing in the line
model = torch.load(fname_model, map_location="cpu")
I think is because i don't have enought ram but i saw (in others repositories on twitter) a dude runned it in an raspberry pi
So... something is not working correctly

[marcuswestin]

Hey @cocktailpeanut if you'd like help managing issues etc I'm happy to help out.

(E.g this one has been fixed and I can mark it if I have permission)