Fix aggregate microbatches bug #711
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This reverts commit 1e08f77.
batchflow/models/torch/base.py
Outdated
| @@ -1551,7 +1551,7 @@ def aggregate_microbatches(self, outputs, chunked_outputs, chunk_sizes, single_o | |||
| for chunk_output, chunk_size in zip(chunked_output, chunk_sizes)], dim=0) | |||
| result.append(output_) | |||
| else: | |||
| result.append(np.mean(chunked_output)) | |||
| result.append(np.mean([chunk_output.item() for chunk_output in chunked_output])) | |||
There was a problem hiding this comment.
The only way we get here (in this method) an instance of torch.Tensor is when the data was explicitly left on device by setting transfer_from_device=False. It is rather strange to return a np.scalar, which is on CPU, in that case
SergeyTsimfer
approved these changes
Aug 1, 2023
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
chunked_output[0].size != 1 always true when passing torch.Tensor(size is a method). Changed it to get_size() function. Also the np.mean function falls, when passing torch tensors with device cuda, so I added .item() to all elements of chunked_output.