hf download: cannot locate the file on the Hub

The message

An error happened while trying to locate the file on the Hub and we cannot find the requested files in the local cache. Please check your connection and try again or make sure your Internet connection is on.

What it means

huggingface_hub could not reach the Hub and has no cached copy. It is a network failure, not a missing repo, and common from mainland China.

Why it happens

• huggingface.co is unreachable from this network: blocked, timing out, or behind a proxy the tool is not using. This is the branch the library takes when the request fails for a reason other than a missing repo, a gated repo or a bad token.
• A gated or private repo is a different message (a 401 or a gated-repo error), so this one is not a token problem.

How to fix it

1. Point the library at a mirror: export HF_ENDPOINT=https://hf-mirror.com before hf download. HF_ENDPOINT is huggingface_hub's own setting for the Hub address; hf-mirror.com is a third-party mirror, not run by Hugging Face. The Chinese pages of this site's command generator already add it.
2. ollama pull hf.co/… does not read HF_ENDPOINT. Download the GGUF with hf download through the mirror, then import it with a Modelfile (FROM ./file.gguf) and ollama create.

Related errors

• Ollama: pull model manifest: file does not exist