Transformers
GGUF
llama

conserving disk space when dealing with split model files

#5
by itsdotscience - opened

Just a little tip to avoid requiring double the space with the split model files. Works on windows as well as *nix and likely macos (dont have that to test atm)

example:

curl "https://huggingface.co/TheBloke/Airoboros-L2-70B-2.1-GGUF/resolve/main/airoboros-l2-70b-2.1.Q6_K.gguf-split-a" -s -L >> airoboros-l2-70b-2.1.Q6_K.gguf

curl "https://huggingface.co/TheBloke/Airoboros-L2-70B-2.1-GGUF/resolve/main/airoboros-l2-70b-2.1.Q6_K.gguf-split-b" -s -L >> airoboros-l2-70b-2.1.Q6_K.gguf

Basically instead of mucking about with the two parts after downloading we're appending each one to a single file. Only downside is It'll make resuming a download more complicated.

Sign up or log in to comment