search

cswry / SeeSR

[CVPR2024] SeeSR: Towards Semantics-Aware Real-World Image Super-Resolution
Apache License 2.0
390 stars 25 forks source link

How many epochs do you set when creating training datasets? #6

Closed XsherryR closed 9 months ago

XsherryR commented 9 months ago

The default setting is 1; is this sufficient for fine-tuning the model?

cswry commented 9 months ago

By default, it is set to 1, but this is definitely not enough for training.

It requires finding a balance between memory consumption and training data diversity.

XsherryR commented 9 months ago

By default, it is set to 1, but this is definitely not enough for training.

It requires finding a balance between memory consumption and training data diversity.

Could you tell me how many epochs you set when creating training datasets?

cswry commented 9 months ago

By default, it is set to 1, but this is definitely not enough for training. It requires finding a balance between memory consumption and training data diversity.

Could you tell me how many epochs you set when creating training datasets?

It could be 60? It has been a long time since the data was created from now.

XsherryR commented 9 months ago

By default, it is set to 1, but this is definitely not enough for training. It requires finding a balance between memory consumption and training data diversity.

Could you tell me how many epochs you set when creating training datasets?

It could be 60? It has been a long time since the data was created from now.

Got it, thanks！