How to prevent repeated downloading with HuggingFace

Question

Description:

I am confused on how the installation of the packages are performed. Currently I was working on a StableDiffusion model and every-time I run the code its again and again downloading files which are 3 to 4 Gigs big.

Code:

This is the code I was trying to run at first:

from torch import autocast
from diffusers import StableDiffusionPipeline

pipe = StableDiffusionPipeline.from_pretrained(
    "CompVis/stable-diffusion-v1-4", 
    use_auth_token=True
).to("cuda")

prompt = "a photo of an astronaut riding a horse on mars"
with autocast("cuda"):
    image = pipe(prompt)["sample"][0]  
    
image.save("astronaut_rides_horse.png")

Issue:

When I run the code the following appears in my shell:

Fetching 16 files:   0%|                                                             | 0/16 [00:00<?, ?it/s]
vae/diffusion_pytorch_model.safetensors:   0%|                                   | 0.00/335M [00:00<?, ?B/s]
unet/diffusion_pytorch_model.safetensors:   0%|                                 | 0.00/3.44G [00:00<?, ?B/s]
safety_checker/model.safetensors:   0%|                                         | 0.00/1.22G [00:00<?, ?B/s]
text_encoder/model.safetensors:   0%|                                            | 0.00/492M [00:00<?, ?B/s]

and this happens each and everytime I run the code.

What I tried?

I tried installing and cloning the whole git repo. (I honestly don't know why I did that even though I know it wasn't gonna affect a thing!) Also I tried searching for many forums for this issue but not even a single clue, maybe its because of my in-experienced approach.

Answer 1

There are 2 possible solutions:

Saving manually:

from diffusers import StableDiffusionPipeline

model = StableDiffusionPipeline.from_pretrained(
    "CompVis/stable-diffusion-v1-4",
    use_auth_token=True,
)

model.save_pretrained("./my_model_directory/")  # only needed first run
model = StableDiffusionPipeline.from_pretrained("./my_model_directory/")

Cache dir:

from diffusers import StableDiffusionPipeline

model = StableDiffusionPipeline.from_pretrained(
    "CompVis/stable-diffusion-v1-4",
    cache_dir="./my_model_directory/",
    use_auth_token=True,
)