Using smolDocling alongside docling
Hi my use case scenario is to use docling for all document parsing all the input it supports, however I want to use smolDocling for unstructured pdfs and images. What would be a viable architecture to do this, currently I was thinking to use the docling library for all documents except pdf and then to host smolDocling on a cloud service and use its endpoint to parse unstructured pdfs and images is that a viable workflow.
@abdbrainy Smoldocling is natively integrated into docling, just update the pipeline settings (see here: https://github.com/docling-project/docling/blob/main/docs/examples/minimal_vlm_pipeline.py#L37)
Error parsing document with VLM: It looks like the config file at '/root/.cache/huggingface/hub/models--ds4sd--SmolDocling-256M-preview/snapshots/492bde898f2bed6b493b4da8256c93de29e03a9b/preprocessor_config.json' is not a valid JSON file.
I get this error while using smoldocling by editing the vlm pipeline.