Is it possible to use a pretrained ResNet50 for SWAV for custom dataset?

[sarmientoj24]

This code block loads a ResNet50 that I assume is not pretrained. Is that correct?

    # build model
    model = resnet_models.__dict__[args.arch](
        normalize=True,
        hidden_mlp=args.hidden_mlp,
        output_dim=args.feat_dim,
        nmb_prototypes=args.nmb_prototypes,
    )

Is it possible that I load torchvision resnet50 here? I am using a different dataset for SWAV.

[irsman06]

Hi @sarmientoj24 ,
Were you able to run swav with a custom dataset?

[Yuxin-Du-Lab]

Is the effect ideal? I'm very interested in this.