Open firmai opened 2 months ago
Thanks for the suggestion! Just to be sure I understand correctly, did you mean using Dynamic Topic Modeling using quantized LLMs?
Yes exactely, I think this is a very interesting research direction. Simple but powerful idea.
Thanks, I understand! It should be doable to implement here since it only requires calculating the representations using the representation models that are already there. Unfortunately, since I work on this in my spare time, I have limited time available to work on these kinds of features. Any help would be appreciated if anybody is interested!
It would be interesting to see if DTM ius possible with quantized models.