wwydmanski / cosine-dim-reduct Goto Github PK

Hi,
Thank you for the repo and the youtube explanation. I am trying out dimensionality reduction for an unsupervised document similarity task. I tried to reproduce your dimensionality reduction methods but could't get around the first step where you use load: "embeddings = np.load("data/vectors.npy")". I generate embeddings for the documents from different methods (GPT3 embeddings, transformer sentence embeddings, pooling, etc). I am not sure how to integrate these embeddings in your code where I try out the different dimensionality reduction methods you have in the notebooks in the repo. Thanks for your help.