Peter M. Elias
Peter M. Elias
Possibly yes. Also could be something about the auto-sizing interfering with the scaling options. This will be fixed soon as I am going to remove that drawing logic entirely since...
Yeah that confirms it's a watch/apply issue though thank you.
I have narrowed down the exact cause of this issue and it will be fixed shortly.
Grammar processing appears to be quite slow (again?): https://github.com/ggerganov/llama.cpp/pull/4306#issuecomment-1947021051
I've noticed it varies widely with respect to prompt complexity. My JSON schema -> grammar contains three levels of object-arrays and if I ask for a shorter output it completes...
Just to close the loop on my previous comment-- I continued experimenting with this feature on a wide variety of cases and ultimately concluded that the performance variance is too...
@alaisi Haha, I ended up discovering all that the hard way :) Ended up creating a dedicated LISTEN service that uses `psycopg2` as it has more mature support for this...
We would like to be able to specify a timeout and also specify the action after timeout. For example: ``` steps: - block: "Deploy to prod?" timeout: after: 20m then:...