Closed VincenDen closed 1 year ago
Hello, In Figure3 T-SNSE of the DM paper, how to extract the features of synthetic images,does it extract features from a model trained on whole dataset or from a model trained on synthetic images? Thanks.
Hi, we extract the features (last-layer logits) with a ResNet trained on whole original dataset.
Hello, In Figure3 T-SNSE of the DM paper, how to extract the features of synthetic images,does it extract features from a model trained on whole dataset or from a model trained on synthetic images? Thanks.