search

yue-zhongqi / ifsl

[NeurIPS 2020] Released code for Interventional Few-Shot Learning
164 stars 23 forks source link

RunTimeError: Tensor size mismatch #11

Open sdustjack opened 2 years ago

sdustjack commented 2 years ago

Hello. I run python main.py --method metatrain --train_aug --test maml5_ifsl_wrn_tiered and got an error like this:

File "E:\LinZD\IFSL\MAML_MN_FT\methods\DMAML.py", line 168, in train_loop loss = self.set_forward_loss(x) File "E:\LinZD\IFSL\MAML_MN_FT\methods\DMAML.py", line 152, in set_forward_loss scores = self.set_forward(x, is_feature=False) File "E:\LinZD\IFSL\MAML_MN_FT\methods\DMAML.py", line 106, in set_forward split_support, support_d, split_query, query_d = self.feature_processor.get_features(support, query) File "E:\LinZD\IFSL\MAML_MN_FT\methods\meta_toolkits.py", line 145, in get_features support_d = self.get_d_feature(support) File "E:\LinZD\IFSL\MAML_MN_FT\methods\meta_toolkits.py", line 139, in get_d_feature d_feature[i] = pd RuntimeError: The expanded size of the tensor (64) must match the existing size (351) at non-singleton dimension 1. Target sizes: [25, 64]. Tensor sizes: [25, 351]

Any help would be appreciated! Thank you!!

My Environment: python 3.7.6, pyTorch 1.4.0 and necessary datasets and packages