Add a warning in gradcheck if inputs precision < float64 #8663
Conversation
This file contains hidden or 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 PR adds a warning to alert users about the possibility of a failure in the gradcheck
vishwakftw
requested review from
apaszke,
colesbury,
ezyang,
gchanan,
soumith and
zdevito
as code owners
ssnl
requested changes
Jun 19, 2018
torch/autograd/gradcheck.py
Outdated
| prec_flag = True | ||
| if any(typecheck): | ||
| warnings.warn( | ||
| 'At least one of the inputs is of single precision. ' |
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
Collaborator
|
Could you update the warning message too? Thanks! |
Contributor
Author
|
Currently, this is the warning message: I could change it to: |
Collaborator
|
What I meant was that the tensor could be half, etc. I think it might be better to read |
petrex
pushed a commit
to petrex/pytorch
that referenced
this pull request
Jun 20, 2018
* upstream/master: (92 commits) more formatting (pytorch#8701) Fix pytorch#8692 (pytorch#8699) Create captured inputs recursively for loop to resolve loop-carried dependencies across nested blocks (pytorch#8345) Shard test_nn to reduce runtime for each test target (pytorch#8678) Create at::tensor (pytorch#8475) Clarify mp note about sharing a tensor's grad field. (pytorch#8688) Add owner rule for cpp_extension.py (pytorch#8700) fix formatting in :math: in fold docstring (pytorch#8696) Some 0-sized dimension support, port catArray away from resizeLegacy. (pytorch#8666) Implement flatten function (pytorch#8578) Created Tensor::to functions (pytorch#8643) Add a warning in gradcheck if inputs precision < float64 (pytorch#8663) Fix parsing of floating point defaults in python_arg_parser (pytorch#8681) Export ProcessGroupGloo options to Python (pytorch#8664) Fix build error in pybind_state_ideep (pytorch#8684) Compatibility: write nDimension/_nDimension corresponding to dim()/_dim(). (pytorch#8676) Improve win-build.sh for local build (pytorch#8674) don't do unnecessary copies for bernoulli_ (pytorch#8682) Use parallel if get_num_threads 0 (pytorch#8677) Fix serialization for Parameters (pytorch#8633) ...
petrex
pushed a commit
to petrex/pytorch
that referenced
this pull request
Jun 21, 2018
* Solves pytorch#8659 This PR adds a warning to alert users about the possibility of a failure in the gradcheck * Fix lint * Update gradcheck.py * Update gradcheck.py * update error message * Update warning message to be more descriptive
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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.
Fixes #8659 . This PR adds a warning to alert users about the possibility of a failure in the
gradcheckandgradgradcheckfunctions.Why warning? Users might alter the values of
eps,atoland/orrtolto make their custom tests pass. Throwing errors in such scenarios could be bad.cc: @ssnl