For round 2: since testset 2 (the testset of round 2) is near to the trainset and testset 1, adding testset 1 data in training would probably help to improve the model performance. However, we do not have testset 1 labels. If one wants to use trainset + testset 1 in training, he should first train a good “round 1” model. This means that if someone wants to succeed in round 2, his round 1 model needs to be the best (or among the bests).
My question is: do you plan to release the ground-truth of testset 1, so that round 2 becomes independent of round 1?
Thank you in advance for your answer.