The LR of the training set has 4 compression qualities, while the validation set only has quality "1". So, which compression quality should be used for the test set? If there are separate compression qualities, can we use 4 models separately?
Posted by: Accusefive @ Feb. 6, 2024, 3:18 a.m.Thank you for participating in our challenge! For the validation phase, we use a mix of compression levels (1 to 4) randomly applied per video to reduce data transfer sizes for submissions. In contrast, the test phase will evaluate on all four compression levels. We are looking for one model that performs well across all compression levels, so in the test phase, participant should submit one model.
Posted by: Ahmed_Telili @ Feb. 6, 2024, 4:30 a.m.