If like me, you want to see what it looks like : https://home.cern/news/news/accelerators/autopsy-lhc-beam-du...

So now I'm reading:

https://iopscience.iop.org/article/10.1088/1748-0221/16/11/P...

And I want microphones, I want to hear the thing ring with that 2.3kHz note. I want to feel the 27Hz wiggle and the 196Hz thump. I want to get the Slow-Mo guys in there to place their camera, and watch the thing jump when a beam hits it.

The amount of energy in that thing just defies intuitive understanding from reading a paper, I have to use other senses.

The amount of materials science knowledge so casually on display is amazing.

You should look at LayoutLM models for a NER task. Then your pipeline should look like : - Identity the menu sub structure (title, item list ...) - Classify each item with 2 labels.

The training process is not hard, but the data gathering / cleaning / labelling can be a little long.

Thanks! I haven't heard of LayoutLM but something that can understand structure from a few examples could be just what I need.

Do one that parse addresses or first name/last name reliably and I'll pay for it.