Manual download of default models

[Arin1599]

I want to use docling in a environment without internet connectivity. I want to manually download the model files in a particular folder and use that path to invoke the docling functionalities . how can I do it? Please Help.

[dosubot[bot]]

Hi @Arin1599! I'm Dosu and I’m helping the docling team.

You can manually download all required Docling models for offline use by running:

docling-tools models download --all -o models

This saves the models in a folder named models. To make Docling use these local models, set the environment variable DOCLING_SERVE_ARTIFACTS_PATH to the path of your models folder when you start Docling. For example:

export DOCLING_SERVE_ARTIFACTS_PATH=/path/to/models

If you're running Docling in a container, mount the models folder and set the same environment variable inside the container. This setup works for both local and production deployments and avoids any runtime downloads, making it suitable for offline environments. More details and examples are available in the official documentation.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}    

[dolfim-ibm]

See https://docling-project.github.io/docling/usage/#model-prefetching-and-offline-usage

[Arin1599]

I understand this, but since I am in a secured company network environment, while connecting to internet for even downloading the mode,
I am facing the following error:

'[SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed: unable to get local issuer certificate (_ssl.c:1002)')))

I can only download the model files from browser and place it in a particular directory and provide the path for it.. Can it be done?

This is the only workaround that I know of..

[cognetics-ai]

Download on your personal machine first. And then upload to your non-work GitHub account using GitHub's large file storage. Github should be accessible inside your enterprise. POC-ing is one thing, taking these frameworks that use 3rd party models all the way to production, will be governed by internal MRM processes.