You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is your feature request related to a problem? Please describe.
When I download a large model via WiFi, sometimes the connection drops and this leads to a complete restart of the download.
Describe the solution you'd like
If the connection is dropped and the download needs to restart, it should continue to download at the point where it stopped, instead of a complete restart.
Describe alternatives you've considered
When you use an external ollama instance you can simply recover interrupted downloads by restarting them at the point where the connection was lost. This is not possible with the integrated ollama instance.
Additional context
I’m using Alpaca on openSUSE tumbleweed as flatpak
The text was updated successfully, but these errors were encountered:
Something similar happens, but the connection is stable. I can't download these unlucky 20 GB. For example, it downloads up to 18%, then suddenly drops back down to 13%. It feels like you're climbing a hill only to slide back down again. In previous versions, I didn't notice this issue even though I was downloading smaller models.
Is your feature request related to a problem? Please describe.
When I download a large model via WiFi, sometimes the connection drops and this leads to a complete restart of the download.
Describe the solution you'd like
If the connection is dropped and the download needs to restart, it should continue to download at the point where it stopped, instead of a complete restart.
Describe alternatives you've considered
When you use an external ollama instance you can simply recover interrupted downloads by restarting them at the point where the connection was lost. This is not possible with the integrated ollama instance.
Additional context
I’m using Alpaca on openSUSE tumbleweed as flatpak
The text was updated successfully, but these errors were encountered: