Feature request: docker image containing createTraining as web API

[cboulanger]

Continuing on from #1167 (comment) :

it would be really great to be able to create training files using the web API via the Docker image. Currently, that service is missing and can only be called via the batch processing, which requires installing and building Grobid.

[lfoppiano]

This is quite useful indeed. I'm trying to get it assembled this week!

[lfoppiano]

Here it is #1309, could you give it a try?

curl --location 'http://localhost:8070/api/createTraining' \
--form 'input=@"/Users/lfoppiano/Downloads/2304.10428v4.pdf"'
--output file.zip

Do you think you could help me adding the documentation for it?

[cboulanger]

Here it is #1309, could you give it a try?

Hi, thanks, should I use lfoppiano/grobid:latest-develop?

Do you think you could help me adding the documentation for it?

I saw you added something already, I am not sure I can contribute anything meaningful but I am happy to if I notice that something is missing.

[cboulanger]

Here it is #1309, could you give it a try?

Hi, thanks, should I use lfoppiano/grobid:latest-develop?

Ah ok seems like that's not the case:

INFO  [2025-07-08 08:23:01,194] io.dropwizard.jersey.DropwizardResourceConfig: The following paths were found for the configured resources:

    POST    /api/annotatePDF (org.grobid.service.GrobidRestService)
    POST    /api/citationPatentAnnotations (org.grobid.service.GrobidRestService)
    GET     /api/grobid (org.grobid.service.GrobidRestService)
...

I'll try the CRF version.

[lfoppiano]

@cboulanger Let me bake you the right image.

[lfoppiano]

@cboulanger you can use lfoppiano/grobid:latest-crf or lfoppiano/grobid:latest-develop

[cboulanger]

It works, thank you very much!

[lfoppiano]

Thanks! It's merged 🙂

[cboulanger]

Now I just need the 100 people who will help me to correct the training data hahaha 😎

[lfoppiano]

Which model(s) do you plan to work on? And how much data do you have to correct, roughly?