galeos

Is there scope to implement ternary models using this approach to minimise die area of the model parameters?

How did you find ModernBERT performance Vs prior BERT models?

My understanding is that BERT can still outperform LLMs for sentiment classification?