Hi all, I have an issue with bool value of tensor with more than one value in this notebook: N835242
I am trying to pass two parameters: indices and offsets into an embeddingbag proxy module (emb3) that is supposed to return a torch.Tensor and the I want to pass the output of this node into the intermediate node along with the same arguments of offsets and per_sample_weights from before. The return of this should be a tuple of Tensors.
Then finally I want to pass the output of the intermediate node into another embedding bag proxy module
When I don’t include offsets or per_sample_weights as input to the emb3 node then I no longer get this bool value of tensor error however my model then no longer recognizes the parameter offset.
I’m not entirely sure why this happens and would appreciate any advice, thank you!