Question regarding ORCA model training history

[Ling Chen]

Hi All,

Orca is trained in multiple steps (1Mb, 1-32Mb, 32 – 256Mb). I wonder what the model’s performance is after each step. For example, after train_h1esc_a (https://github.com/jzhoulab/orca/blob/main/train/train_h1esc_a.py), what is the correlation and mse for the training and validation dataset?

I’m trying to understand how much performance gain on the 1Mb region’s prediction is from incorporating coarse prediction from lower resolution input.

Thank you!

Ling

[Jian Zhou]

Hi Ling,

Not if my previous message was posted to the group. So I am sending it again. The information you are looking for is in Supplementary Table 1 and 2 of the paper.

Best,

Jian