Closed Jack-bo1220 closed 5 months ago
encoder_layer = nn.TransformerEncoderLayer( TypeError: init() got an unexpected keyword argument 'norm_first'
This will show an error after I delete it:
transformer_output = self.transformer_encoder(x) tgt_len, bsz, embed_dim = query.size() ValueError: not enough values to unpack (expected 3, got 2
I found that the shape of x in “transformer_output = self.transformer_encoder(x)” is torch.Size([132, 128])
This seems to be highly related to the torch version, could you tell me the torch version used for the implementation?
I tried 2.0.1+cu117 and it worked smoothly
I got it. thank you so much
encoder_layer = nn.TransformerEncoderLayer( TypeError: init() got an unexpected keyword argument 'norm_first'
This will show an error after I delete it:
transformer_output = self.transformer_encoder(x) tgt_len, bsz, embed_dim = query.size() ValueError: not enough values to unpack (expected 3, got 2
I found that the shape of x in “transformer_output = self.transformer_encoder(x)” is torch.Size([132, 128])