Open lucbouge opened 2 years ago
Hi @lucbouge !
I wish you a very good new year and I hope you're going fine.
Normally you can point to an external grobid home by modifying the configuration file, see https://grobid.readthedocs.io/en/latest/Configuration/#grobid-home
The models under this external grobid-home/models/
will be used.
However, if you work with a fork, updated/push in the fork with different models, you can still normally merge new updates from the original repository without troubles (see for instance https://www.c-sharpcorner.com/article/how-to-merge-upstream-repository-changes-with-your-fork-repository-using-git/). You might have conflicts if the same models are updated in the original repository, but can still manage that manually.
In one of my grobid fork, I have a mechanism to manage custom models (called "flavors") but I didn't find the time yet to merge it - it would also answer your need I think.
Dear Luca, dear Patrice,
I am currently running an experiment at ANR about extracting the publications from final reports of the projects of the last years. I have a batch of more than 400 reports to process for year 2016.
I managed to train Grobid about the specific format of those reports, which is hopefully not that different from a scientific paper My question is now to run the service.
According to the documentation, I started the service by running
./gradlew
and used the Grobid Python client. I works fine, but it uses the models which are physically in the grobid repository, atgrobid-home/models/
. This leads to modify the Git files, which prevents any further pull update.Is there a way of activating the service with some external set of models? Sort of a
--model /path/to/models
option?Regards, Luc.