Junlin Zhou

I'm new on text generation tasks but I want to lower the "creativity" of the model and stick to stable outputs, and according to my limited knowledge I think should...

Thank you for the advice, I will try greedy encoding. For this issue, I mean if nan logprobs means something goes wrong, maybe we can return http status 400 or...

Absolutely, particularly when considering the remarkable performance of the 70B model. Expanding the context length would undoubtedly unlock its full potential, allowing it to excel in tasks such as COT...

I believe the issue lies with your network, preventing access to gcr.io. To resolve this, consider utilizing kustomization rather than modifying the image on the upstream. Here's a kustomization file...

Why is this issue closed? I met a similar situation, is there any solution?

@Sheikh-Abubaker Yes, I think so, I intend to deploy them into a specific namespace, such as placing Loki in the 'logging' namespace. However, I'm uncertain if the absence of namespace...

Thank you for your response!

Hello @chayim I don't know who to pin so I random picked you, and I apologize if this choice is inappropriate. As there has been several PRs trying to upgrade...

@gaby I believe it's the black version (from 23.x to 24.x) that breaks the lint. Upgrading pydantic version has nothing to do with lint.

@gaby I'm sorry I confused poetry version syntax with pipenv, the version of black is limited to 23.x. I have bad memories about upgrading black to 24.x. Also I did...