Help with the "precomputable_affine" Model from Thinc.api

[dhruvbhimani-2512]

I was going through the internal code of spaCy NER model and I came across the file: https://github.com/explosion/spaCy/blob/2eef47dd26a5acbc3f667a2bc3b1ddf16f2d1b07/spacy/ml/_precomputable_affine.py

It is using "precomputable_affine" model which I couldn't find anywhere in Thinc's website. Can anybody help me understand this or share some references?

Thanks.

[thomashacker]

Hello,
the precomputable_affine model is defined in this file and contains the forward- , backpropagation- and initialization function. There are no further docs on this particular model.

[jaskaran-td]

@dhruvbhimani-2512 did u got any idea for precomputable_affine architecture.

[dhruvbhimani-2512]

@jaskaran-td not yet. Please do post here if you have any ref.

[jaskaran-td]

sure man.

[jaskaran-td]

@dhruvbhimani-2512 you can read this blog for the better understanding of ner.

[dhruvbhimani-2512]

Hi @jaskaran-td , this blog is about text parsing, right? Can you explain how is this related to NER?