Does FLASHTransformer accept negative values in sentence embedding?
I passed the BERT sentence embedding into the FLASHTransformer, and got the following error:
IndexError: index out of range in self
I then tried the README method:
x = torch.randint(0, 20000, (1, 1024)) <-- works fine
x = torch.randint(-1, 1, (1, 1024)) <-- same error
Does FLASHTransformer accept negative values in sentence embedding? I passed the BERT sentence embedding into the FLASHTransformer, and got the following error:
I then tried the README method: x = torch.randint(0, 20000, (1, 1024)) <-- works fine x = torch.randint(-1, 1, (1, 1024)) <-- same error
Thx