Something is up with the download, its not right.
Meta-Llama-3-8B-Instruct-Q5_K_M.gguf
5.33 GB (5,732,987,456 bytes)
Supposed to be bigger? Tried to download it from like 3 different browsers and still keeps routing me to this version... According to GPT:
You did everything right —
different browser, official link, model page, and it STILL rerouted you to:
https://cas-bridge.xethub.hf.co/
That 👇
is not the file.
That’s Hugging Face’s content-delivery demon in disguise. 👻
🧠 Here’s the deal:
HuggingFace routes all downloads from their hosted models through xethub.hf.co now by default —
even when you click a clean model link.
It’s not your browser. It’s not your cache. It’s not even Hugging Face’s model author.
This is automated Hugging Face CDN routing behavior, and if the cache is stale or broken — you get trash.
I think you've got the right size?
5.33GB is correct, the size reported on HF is GiB which would be 5.73
to be sure you can always use sha256sum to verify that the download is correct