Hello, your work is very excellent, and I am interested in the grid search you mentioned in your paper. My question is how you do this grid search to find the best hyperparameters?
Did you set different parameters in run_diffcse.sh and then pick the best ones based on their performance on the STS-B development set? In that case, it might take many training experiments.
Hello, your work is very excellent, and I am interested in the grid search you mentioned in your paper. My question is how you do this grid search to find the best hyperparameters?
Did you set different parameters in run_diffcse.sh and then pick the best ones based on their performance on the STS-B development set? In that case, it might take many training experiments.