SpeedUp with microsoft/nni

[zhiqwang]

🚀 The feature

microsoft/nni#5345

Motivation, pitch

To be updated

Alternatives

No response

Additional context

No response

[chandan-wiai]

Taking the discussion forward from here.

Does this mean, I can take my checkpoint trained on yolort.models.YOLOv5 and load as shown above, and that model object won't have 'transform' module?

Hi @chandan-wiai , I guess not in this scenario. There are no parameters or buffers in YOLOTransform modules, it should be easy theoretically. Maybe we should build the model as following from this api:

from yolort.models.yolo import yolov5_darknet_pan_s_r60  # aka yolov5s

model = yolov5_darknet_pan_s_r60()  # we do not specify pretrained=True, i.e. do not load default weights
model.load_state_dict(torch.load('checkpoint_from_yolort.pt'))
model.eval()

We can also discuss this ticket at #484 so as not to disturb more people for questions not related to nni.

Got it. Basically directly doing this step. I think this should help. Thanks.

[chandan-wiai]

BTW @zhiqwang , why return a nested tensor as it is done here because I don't see samples.image_sizes being used anywhere?

[zhiqwang]

Hi @chandan-wiai , NestedTensor can be removed I guess, I didn't do it because I'm working on other projects now. See #471 for more details, and contributions are welcome here.