Finetuning QE models.
#3
by
dchaplinsky
- opened
Hi @jmprcp .
I'm evaluating the idea to finetune the existing comet models for some of the pairs of my interests. Any guidance on the number of sentences in DA datasets required for a training?
Hi! You can check out our papers on COMET and subsequent models for details on this: COMET, COMET-22, COMETKiwi-22, COMETKiwi-23, and xCOMET.
jmprcp
changed discussion status to
closed