fix: use torch.from_numpy()
instead of torch.Tensor()
to keep data type
#2951
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Outdated suggestions cannot be applied.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.
When running pytorch runner with
async_run
method in production mode(not in dev mode), tensors with other type thanFloat32
type is converted intoFloat32
type.(Ex.LongTensor
is converted toFloatTensor
) And I think this bug is originated by this line.I made small test to check this issue. (numpy == 1.22.3 / torch == 1.12.1)
Thus I think changing
torch.Tensor(ret)
totorch.from_numpy(ret)
should resolve this bug. Please tell me if it was a intended behavior.(i.e converting every tensor intoFloat32
tensor). Thank you!