Model Architecture

#15
by ovska - opened

Hello,
Thank you for the journey into complex space that your work has taken me on.
I do have a few questions about this UAE model.

  1. Was the AnglE fine-tuning done on BERT or LLaMA for this model, or something else?
  2. What is the pooling strategy used?
  3. What was the data used for the fine-tuning? Is it just the LLM generated labelled data you mention in your paper or other datasets?
    Thank you!

Sign up or log in to comment