So it just occurred to me that the recommended approach for training new vectors is to use the en_core_web_sm model or a blank model like what @honnibal mentions in this post
I’m going to retrain the vectors using this model and see if that solves my issue.
For additional context. If I import the model using Spacy and attempt to process text I get the same dimension error. If however, I disable the parser, tagger, and ner components while loading my custom model, I’m able to utilize the rest of the language model without the dimension issue.