Hey all! Can I use Kern AI for sequence tagging?

What open-source datasets are there that would also have a good sentence-encoder on hugging face?

@mention If you're aiming to open-source something, let us know in any way we can help. We've also talked to other awesome OS startups, and that helped a ton 🙂

What was /where is the article you guys wrote about how to do a release 'well' to get attention from the open-source community?

Or @mention , are you referring to using the application with a single account while sharing the credentials with the others? That would work with the OSS refinery! (you would just lose the multi-user integrations like inter-annotator agreement and gold-star labeling etc.)

Hello! Thank you for accepting me in the group. I recently discovered Kern Refinery and I am trying to set it up on a remote server for 2-3 users including myself. My problem is that I cannot access the UI remotely from a browser. I have tried forwarding port 4455 but I get a 404 in return. I start all services using the default docker-compose.yml by running ./start, as described on github README. Has any body else had the same issue ? Any thoughts?

Hi @mention, thanks for the help with the issue on GitHub. I've some more specific questions, wondered if you may help. I have some ; separated CSVs that I'd like to use in Kern and I'm not really sure how to specify this. I've tried with `sep=;` and `sep=';'` as stated in the pandas docs, but it seems to keep looking for `,` to divide fields. For example, for a simple file like this: ``` a;b;c 1;2;hola ``` The data schema looks like this (data-schema.png). How should I pass the parameter?. Thanks in advance

I don't know how to tweak the dockerfile to enable others to hit my IP and quickly view it? Normally involves specifying the IP to be 0.0.0.0 somewhere instead of localhost

And it always take me a while to find the place to delete projects

Also, happy to write this somewhere more formally, but it's very easy to loose the way back to the main projects page, the kern logo doesn't look like a button quite enough for the design to work