Fixes to make life easier with the nlp library #6423
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
Codecov Report
@@ Coverage Diff @@
## master #6423 +/- ##
==========================================
+ Coverage 77.51% 79.79% +2.27%
==========================================
Files 150 150
Lines 27789 27807 +18
==========================================
+ Hits 21542 22188 +646
+ Misses 6247 5619 -628
Continue to review full report at Codecov.
|
sgugger
commented
Aug 11, 2020
| @@ -2318,7 +2318,7 @@ def _concat_inputs_history(self, inputs: List[List[int]], histories: List[Option | |||
| max_len = max([len(item) for item in outputs]) | |||
| outputs = [output + [self.pad_token_id] * (max_len - len(output)) for output in outputs] | |||
| outputs = BatchEncoding( | |||
| {"input_ids": outputs, "attention_mask": [1] * len(outputs)}, tensor_type=self.framework | |||
| {"input_ids": outputs, "attention_mask": [[1] * len(outputs)]}, tensor_type=self.framework, | |||
There was a problem hiding this comment.
This is the only thing were the change of dim in BatchEncoding.convert_to_tensors is breaking something, but in this case, it was a bit magical that the dimension was automatically added, so I don't think this is a serious failure.
LysandreJik
approved these changes
Aug 12, 2020
|
Merging then, we can follow up next week when @thomwolf is back if he has more comments. |
fabiocapsouza
added a commit
to fabiocapsouza/transformers
that referenced
this pull request
Nov 15, 2020
)" This reverts commit 79dd29d.
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Outdated suggestions cannot be applied.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.
This PR adds two things to make the interface easier with the
nlplibrary:BatchEncodingstops enforcing a 2-dim for every tensor, which causes problems for labels (which should be one vector of shape[batch_size]).PreTrainedTokenizerBase.padaccepts tensors as inputs, which makes it easy to use this function for data collation.Added proper documentation and tests from @thomwolf initial work.