OSError: We couldn't connect to 'https://huggingface.co'

[alirezaomneky]

Hi all,

I have pushed the weights of my Dreambooth model to Huggingface (I used LoRa model). I confirm that pytorch_lora_weights.safetensors is indeed on Huggingface. When I want to get inference from the model, I run the following (as per the Readme file):

`from huggingface_hub.repocard import RepoCard
from diffusers import DiffusionPipeline
import torch

lora_model_id = <"lora-sdxl-dreambooth-id">
card = RepoCard.load(lora_model_id)
base_model_id = card.data.to_dict()["base_model"]

pipe = DiffusionPipeline.from_pretrained(base_model_id, torch_dtype=torch.float16)
pipe = pipe.to("cuda")
pipe.load_lora_weights(lora_model_id)
image = pipe("A picture of a sks dog in a bucket", num_inference_steps=25).images[0]
image.save("sks_dog.png")`

But I keep getting this error:

OSError: We couldn't connect to 'https://huggingface.co' to load this model, couldn't find it in the cached files and it looks like "path" is not the path to a directory containing a file named pytorch_lora_weights.safetensors or
Checkout your internet connection or see how to run the library in offline mode at 'https://huggingface.co/docs/diffusers/installation#offline-mode'.

Can you please let me know how to resolve this issue?

[sayakpaul]

Can you try now? Make sure the internet connection is stable. Or if you don't want to rely on that, download the files once and use the offline mode.

[alirezaomneky]

Can you try now? Make sure the internet connection is stable. Or if you don't want to rely on that, download the files once and use the offline mode.

Hi @sayakpaul , I solved the problem by using the following:

!pip install -q diffusers transformers accelerate peft

It appears that when the weights are stored in LFS format on Huggingface, we can't simply use the path to the file.

[sayakpaul]

Closing the issue then :) Glad that you were able to resolve the problem.

[swerizwan]

https://blog.csdn.net/m0_46295727/article/details/133221439

[askerlee]

I also encountered this error. Turned out I accidentally set TRANSFORMERS_OFFLINE to 1. Setting it to 0 solves the issue.

[premmotgi]

I also encountered this error. Turned out I accidentally set TRANSFORMERS_OFFLINE to 1. Setting it to 0 solves the issue.
Thanks a ton! This fixed my issue

[pavoltravnik]

When i generated token i created wrong type
Create a new access token Type Read instead of default Fine-grained (custom)

[MHRDYN7]

When i generated token i created wrong type Create a new access token Type Read instead of default Fine-grained (custom)

Yes, and the "Write" type also works

[jsale017]

I also encountered this error. Turned out I accidentally set TRANSFORMERS_OFFLINE to 1. Setting it to 0 solves the issue.

How were you able to change this?

[azharlightelligence]

You can set hf mirror to fix the huggingface.co internet connection problem.
Try this:

export HF_ENDPOINT=https://hf-mirror.com

[FRAGGERR]

Try this one:
!pip install -U transformers

import torch
from transformers import AutoModel, AutoTokenizer, BertTokenizer

torch.set_grad_enabled(False)

Store the model we want to use

MODEL_NAME = "bert-base-cased"

We need to create the model and tokenizer

model = AutoModel.from_pretrained(MODEL_NAME)
tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)

[Lareina2441]

您可以设置 hf 镜像来解决 huggingface.co 互联网连接问题。 试试这个：

export HF_ENDPOINT=https://hf-mirror.com

thanks， this works

[marinarosell]

When i generated token i created wrong type Create a new access token Type Read instead of default Fine-grained (custom)

This fixed the issue for me!!

[jayquincey]

When i generated token i created wrong type Create a new access token Type Read instead of default Fine-grained (custom)

Dude, you're a frikkin genius! Would never have guessed my read-only token would work but my fine-grained dev token with read and write would result in connection error!!

Thanks a million.