scavru
scavru
try --no-cache, didn't help.
It worked, but it's not right. KoboldAI works without such manipulation.
because there goes summarizer, and it works so crooked that we end up with an initial query all the time.
it would be nice to disable summarizer and insert the raw promt, the result would be better. I do not know how to do it, you should look here aiserver.py...
I run `There's a prebuilt Docker image with dependencies included at [rainevi/tgreddit](https://hub.docker.com/repository/docker/rainevi/tgreddit).` I add redacted https://github.com/raine/tgreddit/blob/master/config.example.toml in environment docker container config.toml
We waiting...