Thanks for the response! Let's say if I build a pre-trained model using JW3000 dataset. I think that should be made public? Below is the text from rules section.

"You may use pretrained models as long as they are openly available to everyone."

That's not the case in this particular competition. Since it's stated in the rules that: "You may also use the JW300 parallel dataset to augment the data."

So everyone is free to use that dataset in any way that suit him well.

Yes, makes sense. I hope @Zindi will comment if they think otherwise. Thanks Muhamed!

My pleasure !

I don't think you can - better check with @zindi. You can use the parallel dataset to augment your data but pretraining models are banned (most probably due to the GPU resource requirement that not all of us have). Plus you would need to release the pre-trained model before some certain date if you would want to use it

Yes, Zindi just updated the rules saying we can't use JW3000 dataset.