[DataPipe] Adding BatcherMapDataPipe #68197
Closed
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
[ghstack-poisoned]
CI Flow Status⚛️ CI FlowRuleset - Version:
You can add a comment to the PR and tag @pytorchbot with the following commands: # ciflow rerun, "ciflow/default" will always be added automatically
@pytorchbot ciflow rerun
# ciflow rerun with additional labels "-l <ciflow/label_name>", which is equivalent to adding these labels manually and trigger the rerun
@pytorchbot ciflow rerun -l ciflow/scheduled -l ciflow/slowFor more information, please take a look at the CI Flow Wiki. |
🔗 Helpful links
💊 CI failures summary and remediationsAs of commit d825c1c (more details on the Dr. CI page):
🕵️ 1 new failure recognized by patternsThe following CI failures do not appear to be due to upstream breakages:
|
cc @VitalyFedyunin ejguan @NivekT [ghstack-poisoned]
cc @VitalyFedyunin ejguan @NivekT [ghstack-poisoned]
cc @VitalyFedyunin ejguan @NivekT [ghstack-poisoned]
|
@NivekT has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator. |
cc @VitalyFedyunin ejguan @NivekT Differential Revision: [D32440963](https://our.internmc.facebook.com/intern/diff/D32440963) [ghstack-poisoned]
|
@NivekT has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator. |
ejguan
approved these changes
Nov 17, 2021
Comment on lines
64
to
72
| if self.length is not None: | ||
| return self.length | ||
| if isinstance(self.datapipe, Sized) and self.unbatch_level == 0: | ||
| if self.drop_last: | ||
| self.length = len(self.datapipe) // self.batch_size | ||
| else: | ||
| self.length = (len(self.datapipe) + self.batch_size - 1) // self.batch_size | ||
| return self.length | ||
| raise TypeError("{} instance doesn't have valid length".format(type(self).__name__)) |
There was a problem hiding this comment.
IMO, we should assume __len__ is always implemented for MapDataPipe
There was a problem hiding this comment.
I agree, though there are cases where it can be hard to compute (especially when one element in the input DataPipe can expands into many).
There was a problem hiding this comment.
The unbatch_level is super annoying here. Otherwise, we don't need to check if self.datapipe is Sized.
There was a problem hiding this comment.
I think you still need to check for it before you call len(self.datapipe) right?
In either case, maybe we should get rid of unbatch_level? I feel that users can call .unbatch prior to this if they wanted that operation. I decided to include it only because IterDataPipe's version has that argument and I hesitate to make the two different.
There was a problem hiding this comment.
If we assume prior datapipe has __len__, then we can directly call len(self.datapipe.
In either case, maybe we should get rid of unbatch_level? I feel that users can call .unbatch prior to this if they wanted that operation. I decided to include it only because IterDataPipe's version has that argument and I hesitate to make the two different.
Yeah. I feel like unbatch_level making me confused even for IterDataPipe. At least based on my experience with Vision and Text, no one really uses it.
There was a problem hiding this comment.
If that is the case, I think we should remove it for simplicity. I will make a PR and let the domains users have a quick look. I agree that it is unlikely that people are using it.
Implements part of #57031 cc @VitalyFedyunin ejguan @NivekT Differential Revision: [D32440963](https://our.internmc.facebook.com/intern/diff/D32440963) [ghstack-poisoned]
|
@NivekT has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator. |
Implements part of #57031 cc @VitalyFedyunin ejguan @NivekT Differential Revision: [D32440963](https://our.internmc.facebook.com/intern/diff/D32440963) [ghstack-poisoned]
|
@NivekT has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator. |
Implements part of #57031 cc @VitalyFedyunin ejguan @NivekT Differential Revision: [D32440963](https://our.internmc.facebook.com/intern/diff/D32440963) [ghstack-poisoned]
|
@NivekT has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator. |
Implements part of #57031 cc @VitalyFedyunin ejguan @NivekT Differential Revision: [D32440963](https://our.internmc.facebook.com/intern/diff/D32440963) [ghstack-poisoned]
|
@NivekT has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator. |
NivekT
added a commit
that referenced
this pull request
Dec 1, 2021
… removing 'unbatch_level' argument and functionality" Based on my conversation with ejguan [here](#68197 (review)), we both believe that having the `unbatch_level` argument and functionality is making this DataPipe unnecessarily complicated, because users can call `.unbatch` before `.batch` if they would like to do so. That will likely be cleaner as well. I also checked other libraries (for example, [TensorFlow](https://www.tensorflow.org/api_docs/python/tf/data/Dataset#unbatch)), and I do not see them provide the ability the `unbatch` within the `batch` function either. This PR simplifies the DataPipe by removing the argument. cc @VitalyFedyunin ejguan @NivekT Differential Revision: [D32532594](https://our.internmc.facebook.com/intern/diff/D32532594) [ghstack-poisoned]
NivekT
added a commit
that referenced
this pull request
Dec 1, 2021
…ch_level' argument and functionality" Based on my conversation with ejguan [here](#68197 (review)), we both believe that having the `unbatch_level` argument and functionality is making this DataPipe unnecessarily complicated, because users can call `.unbatch` before `.batch` if they would like to do so. That will likely be cleaner as well. I also checked other libraries (for example, [TensorFlow](https://www.tensorflow.org/api_docs/python/tf/data/Dataset#unbatch)), and I do not see them provide the ability the `unbatch` within the `batch` function either. This PR simplifies the DataPipe by removing the argument. cc @VitalyFedyunin ejguan @NivekT Differential Revision: [D32532594](https://our.internmc.facebook.com/intern/diff/D32532594) [ghstack-poisoned]
facebook-github-bot
pushed a commit
that referenced
this pull request
Dec 2, 2021
…rgument and functionality (#68594) Summary: Pull Request resolved: #68594 Based on my conversation with ejguan [here](#68197 (review)), we both believe that having the `unbatch_level` argument and functionality is making this DataPipe unnecessarily complicated, because users can call `.unbatch` before `.batch` if they would like to do so. That will likely be cleaner as well. I also checked other libraries (for example, [TensorFlow](https://www.tensorflow.org/api_docs/python/tf/data/Dataset#unbatch)), and I do not see them provide the ability the `unbatch` within the `batch` function either. This PR simplifies the DataPipe by removing the argument. cc VitalyFedyunin ejguan NivekT Test Plan: Imported from OSS Reviewed By: ejguan Differential Revision: D32532594 Pulled By: NivekT fbshipit-source-id: 7276ce76ba2a3f207c9dfa58803a48e320adefed
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.
Stack from ghstack:
Implements part of #57031
cc @VitalyFedyunin @ejguan @NivekT
Differential Revision: D32440963