NVIDIA / modulus

Open-source deep-learning framework for building, training, and fine-tuning deep learning models using state-of-the-art Physics-ML methods
https://developer.nvidia.com/modulus
Apache License 2.0
791 stars 165 forks source link

[CorrDiff]: Only pass interface_kwargs into ResLoss constructor, but empty for EDMLossSR #546

Closed DavidLandup0 closed 1 week ago

DavidLandup0 commented 3 weeks ago

Modulus Pull Request

Description

The PR sets theinterface_kwargs to {} if the loss type is set to modulus.metrics.diffusion.EDMLossSR in CorrDiff's training_loop.py, and uses the existing kwargs for modulus.metrics.diffusion.ResLoss. The kwargs being passed into it are:

interface_kwargs = dict(
            regression_net=net_reg,
            img_shape_x=img_shape_x,
            img_shape_y=img_shape_y,
            patch_shape_x=patch_shape_x,
            patch_shape_y=patch_shape_y,
            patch_num=patch_num,
            hr_mean_conditioning=hr_mean_conditioning,
        )

Checklist

/cc @mnabian

mnabian commented 1 week ago

/blossom-ci

mnabian commented 1 week ago

/blossom-ci

mnabian commented 1 week ago

/blossom-ci

mnabian commented 1 week ago

/blossom-ci

mnabian commented 1 week ago

/blossom-ci

mnabian commented 1 week ago

/blossom-ci

mnabian commented 1 week ago

/blossom-ci