Patrick Devine
Patrick Devine
@olekse it's the same.
We've had a number of different performance fixes to the ollama server since January. I'm going to go ahead and close this, but please feel free to reopen it if...
There's actually only the `-h` flag for `ollama list` right now. Was there anything in particular that you were looking for?
@shivrajjadhav733 are you behind some kind of firewall? Can you `ping registry.ollama.ai`? It looks like DNS resolved correctly.
The `bad request` happens because you're not setting the headers correctly for the registry to understand. That's expected behaviour. To get this to work behind a proxy, you can run`HTTPS_PROXY=...
@shivrajjadhav733 it looks like you're using an `http` proxy and not an `https` proxy with the `HTTPS_PROXY` env variable.
This is actually _somewhat_ covered in the [FAQ](https://github.com/ollama/ollama/blob/main/docs/faq.md#setting-environment-variables-on-linux). The short answer if you want to only configure this for ollama is: ``` $ sudo systemctl edit ollama ``` You'll need...
@ThatOneCalculator each model is pulled synchronously, however, the "chunks" of that model are actually pulled asynchronously, which is how we get fast pull times.
@ThatOneCalculator Given how we're already pulling things, I'm not sure that would help a lot. What speeds are you seeing when pulling right now?
@KhimairaCrypto the etags implementation. It needs to be revamped.