This repository has been archived by the owner on Dec 16, 2022. It is now read-only.
Add influence functions to interpret module #4988
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
…ting instnaces to json files
…into simple-interpreter-new
Submit to the wrong branch
wrong branch
not sure why it's included
not sure why it's included
epwalsh
reviewed
Mar 24, 2021
|
Looks like you have an empty module: |
|
Yeah sure; I am not sure why I kept letting files/folders lurk into my commit :-( |
|
@leo-liuzy looks like you already lost access to Slack 😭 This is looking great! I'll give it a final look on Monday. It was great having you as an intern, and good luck next quarter! |
|
Thanks Pete for being such a great mentor! |
|
TODO:
|
epwalsh
changed the title
epwalsh
approved these changes
Apr 14, 2021
| The top `k` most influential training instances along with their influence score. | ||
| """ | ||
|
|
||
| bottom_k: List[TrainInstanceInfluence] |
There was a problem hiding this comment.
I would expect the bottom k to be basically noise. Not that it's a problem to have this, but you might want to at least warn people in the docstring that this is only questionably useful for most methods.
There was a problem hiding this comment.
Yea, that's a good point. I don't see the value of keeping it, so I'm going to remove.
| """ | ||
|
|
||
|
|
||
| class TrainInstance(NamedTuple): |
There was a problem hiding this comment.
This name is a bit confusing to me. It doesn't immediately say why it's different from Instance. Maybe instead use InstanceWithGrads?
| from allennlp.nn.util import move_to_device | ||
|
|
||
|
|
||
| class TrainInstanceInfluence(NamedTuple): |
There was a problem hiding this comment.
Not sure that Train really adds much to this name. InstanceInfluence is likely sufficient, and already implies that it's a training instance.
Comment on lines
61
to
62
| A `SaliencyInterpreter` interprets an AllenNLP Predictor's outputs by assigning an influence | ||
| score to each training instance with respect to each test input. |
There was a problem hiding this comment.
Suggested change
| A `SaliencyInterpreter` interprets an AllenNLP Predictor's outputs by assigning an influence | |
| score to each training instance with respect to each test input. | |
| An `InfluenceInterpreter` interprets an AllenNLP Predictor's outputs by finding the training | |
| instances that had the most influence on the prediction for each test input. |
| # Load training instances, compute loss and gradients. | ||
| self.train_instances: List[TrainInstance] = [] | ||
| self._model.train() | ||
| for instance in Tqdm.tqdm( |
There was a problem hiding this comment.
Having all of this computation in a constructor makes me really nervous. Maybe it's fine, I just try to avoid this design. We handled this kind of initialization in hotflip by having a method that gets called the first time it's needed.
|
|
||
| # Parameters | ||
|
|
||
| model : `Model`, required |
There was a problem hiding this comment.
Up above you said that this was for predictors, but here's it's a model.
There was a problem hiding this comment.
Yes good point, I'll remove mention of predictors.
| outputs.append(TrainInstanceInfluence(instance=instance, loss=loss, score=score.item())) | ||
| return outputs | ||
|
|
||
| def calculate_influence_scores( |
There was a problem hiding this comment.
This looks like it should be private? I get that you want people writing new InfluenceInterpreters to see this and know they have to implement this method, but you don't want people using an InfluenceInterpreter to call this method, or even have to think about it. Having it here with a public name makes me as a user spend time wondering if I should be using this.
There was a problem hiding this comment.
Yea, I mainly wanted to make sure this showed up in the docs, but I forgot we actually have a way of allowing certain private methods to appear in the docs (like we do with DatasetReader._read()). I'll update to make this private and include in the docs.
epwalsh
added a commit
that referenced
this pull request
Apr 22, 2021
* creating a new functionality to fields and instances to support outputing instnaces to json files * creating tests for the new functionality * fixing docs * Delete __init__.py * Delete influence_interpreter.py * Delete use_if.py * Delete simple_influence_test.py * fixing docs * finishing up SimpleInfluence * passing lint * passing format * making small progress in coding * Delete fast_influence.py Submit to the wrong branch * Delete faiss_utils.py wrong branch * Delete gpt2_bug.py not sure why it's included * Delete text_class.py not sure why it's included * adding test file * adding testing files * deleted unwanted files * deleted unwanted files and rearrange test files * small bug * adjust function call to save instance in json * Update allennlp/interpret/influence_interpreters/influence_interpreter.py Co-authored-by: Evan Pete Walsh <epwalsh10@gmail.com> * Update allennlp/interpret/influence_interpreters/influence_interpreter.py Co-authored-by: Evan Pete Walsh <epwalsh10@gmail.com> * Update allennlp/interpret/influence_interpreters/influence_interpreter.py Co-authored-by: Evan Pete Walsh <epwalsh10@gmail.com> * move some documentation of parameters to base class * delete one comment * delete one deprecated abstract method * changing interface * formatting * formatting err * passing mypy * passing mypy * passing mypy * passing mypy * passing integration test * passing integration test * adding a new option to the do-all function * modifying the callable function to the interface * update API, fixes * doc fixes * add `from_path` and `from_archive` methods * fix docs, improve logging * add test * address @matt-gardner's comments * fixes to documentation * update docs Co-authored-by: Evan Pete Walsh <epwalsh10@gmail.com> Co-authored-by: Evan Pete Walsh <petew@allenai.org>
epwalsh
added a commit
that referenced
this pull request
Apr 22, 2021
* Formatting * New activation functions * Makes position embeddings optional in the transformer embeddings * Adds T5 * Various fixes to make this start up * Share weights * Adds one test that passes, and one test that fails * use min_value_of_dtype in apply_mask * fixes, add beam search * encoder fixes * fix * fix beam search * fix tests * rename to just 'T5' * fix initialization from pretrained * add Model, DatasetReader, and Predictor * remove useless dataset reader * move high-level peices to allennlp-models * revert predictor changes * remove unneeded hidden_size * remove stray comment * bool masks * CHANGELOG * fix test file name * revert other change * revert other change * Distributed training with gradient accumulation (#5100) * Fixes distributed training with gradient accumulation * Fix in case we don't do anything in a batch group * Test for the problematic condition * Formatting * More formatting * Changelog * Fix another test * Fix even more tests * Fixes one more test * I can fix these tests all day. * Add link to gallery and demo in README (#5103) * Add link to gallery in README * Update README.md * try emojis Is this overkill? * Adding a metadata field to the basic classifier (#5104) * Adding metadata parameter to BasicClassifier * Fix * Updating the changelog * reformatting * updating parameter type * fixing import Co-authored-by: Dirk Groeneveld <dirkg@allenai.org> * additional W&B params (#5114) * additional W&B params * add wandb_kwargs * fix * fix docs * Add eval_mode argument to pretrained transformer embedder (#5111) * Add eval_mode argument to pretrained transformer embedder * Edit changelog entry * Lint * Update allennlp/modules/token_embedders/pretrained_transformer_embedder.py * Apply suggestions from code review Co-authored-by: Evan Pete Walsh <epwalsh10@gmail.com> Co-authored-by: Evan Pete Walsh <petew@allenai.org> * specify 'truncation' to avoid transformers warning (#5120) * specify 'truncation' to avoid transformers warning * Update docs * Remove `stride` param * Update CHANGELOG.md Co-authored-by: Dirk Groeneveld <dirkg@allenai.org> * Predicting with a dataset reader on a multitask model (#5115) * Create a way to use allennlp predict with a dataset and a multitask model * Fix type ignoration * Changelog * Fix to the predictor * fix bug with interleaving dataset reader (#5122) * fix bug with interleaving dataset reader * more tests * Update allennlp/data/dataset_readers/interleaving_dataset_reader.py * Update allennlp/data/dataset_readers/interleaving_dataset_reader.py * remove jsonpickle from dependencies (#5121) Co-authored-by: Dirk Groeneveld <dirkg@allenai.org> * Update docstring for basic_classifier (#5124) * improve error message from Registrable class (#5125) Co-authored-by: Akshita Bhagia <akshita23bhagia@gmail.com> * Prepare for release v2.3.0 * fix docs CI * Take the number of runs in the test for distributed metrics (#5127) * Take the number of runs in the test for distributed metrics * Changelog * Add influence functions to interpret module (#4988) * creating a new functionality to fields and instances to support outputing instnaces to json files * creating tests for the new functionality * fixing docs * Delete __init__.py * Delete influence_interpreter.py * Delete use_if.py * Delete simple_influence_test.py * fixing docs * finishing up SimpleInfluence * passing lint * passing format * making small progress in coding * Delete fast_influence.py Submit to the wrong branch * Delete faiss_utils.py wrong branch * Delete gpt2_bug.py not sure why it's included * Delete text_class.py not sure why it's included * adding test file * adding testing files * deleted unwanted files * deleted unwanted files and rearrange test files * small bug * adjust function call to save instance in json * Update allennlp/interpret/influence_interpreters/influence_interpreter.py Co-authored-by: Evan Pete Walsh <epwalsh10@gmail.com> * Update allennlp/interpret/influence_interpreters/influence_interpreter.py Co-authored-by: Evan Pete Walsh <epwalsh10@gmail.com> * Update allennlp/interpret/influence_interpreters/influence_interpreter.py Co-authored-by: Evan Pete Walsh <epwalsh10@gmail.com> * move some documentation of parameters to base class * delete one comment * delete one deprecated abstract method * changing interface * formatting * formatting err * passing mypy * passing mypy * passing mypy * passing mypy * passing integration test * passing integration test * adding a new option to the do-all function * modifying the callable function to the interface * update API, fixes * doc fixes * add `from_path` and `from_archive` methods * fix docs, improve logging * add test * address @matt-gardner's comments * fixes to documentation * update docs Co-authored-by: Evan Pete Walsh <epwalsh10@gmail.com> Co-authored-by: Evan Pete Walsh <petew@allenai.org> * Update CONTRIBUTING.md (#5133) * Update CONTRIBUTING.md * updated changelog Co-authored-by: Akshita Bhagia <akshita23bhagia@gmail.com> Co-authored-by: Arjun Subramonian <arjuns@ip-192-168-0-106.us-west-2.compute.internal> * fix #5132 (#5134) * fix * Prepare for release v2.3.1 * Fairness Metrics (#5093) * Added three definitions of fairness * Updated CHANGELOG * Added DemographicParityWithoutGroundTruth and finished tests * finished refactoring Independence, Separation, and Sufficiency to accumulate * added distributed functionality to Independence, Sufficiency, and Separation * Finished aggregate and distributed functionality for DemographicParityWithoutGroundTruth * fixed GPU and doc issues * fixed GPU and doc issues * fixed GPU and doc issues * fixed GPU issues * fixed GPU issues * added init file * fixed typo * minor docstring changes * minor changes to docstring * Added simple explanations of fairness metrics to docstrings * Further vectorized all metric implementations * Fixed device issue Co-authored-by: Arjun Subramonian <arjuns@Arjuns-MacBook-Pro.local> Co-authored-by: Akshita Bhagia <akshita23bhagia@gmail.com> Co-authored-by: Dirk Groeneveld <dirkg@allenai.org> * fix cached_path for hub downloads (#5141) * fix cached_path for hub downloads * fix test name * fix type hint * Update allennlp/common/file_utils.py Co-authored-by: Lysandre Debut <lysandre@huggingface.co> Co-authored-by: Lysandre Debut <lysandre@huggingface.co> * fix * fix Co-authored-by: epwalsh <epwalsh10@gmail.com> Co-authored-by: Evan Pete Walsh <petew@allenai.org> Co-authored-by: Jacob Morrison <jacob1morrison@gmail.com> Co-authored-by: Nelson Liu <nelson-liu@users.noreply.github.com> Co-authored-by: Akshita Bhagia <akshita23bhagia@gmail.com> Co-authored-by: Leo Liu <zeyuliu2@uw.edu> Co-authored-by: ArjunSubramonian <arjun.subramonian@gmail.com> Co-authored-by: Arjun Subramonian <arjuns@ip-192-168-0-106.us-west-2.compute.internal> Co-authored-by: Arjun Subramonian <arjuns@Arjuns-MacBook-Pro.local> Co-authored-by: Lysandre Debut <lysandre@huggingface.co>
dirkgr
pushed a commit
that referenced
this pull request
May 10, 2021
* creating a new functionality to fields and instances to support outputing instnaces to json files * creating tests for the new functionality * fixing docs * Delete __init__.py * Delete influence_interpreter.py * Delete use_if.py * Delete simple_influence_test.py * fixing docs * finishing up SimpleInfluence * passing lint * passing format * making small progress in coding * Delete fast_influence.py Submit to the wrong branch * Delete faiss_utils.py wrong branch * Delete gpt2_bug.py not sure why it's included * Delete text_class.py not sure why it's included * adding test file * adding testing files * deleted unwanted files * deleted unwanted files and rearrange test files * small bug * adjust function call to save instance in json * Update allennlp/interpret/influence_interpreters/influence_interpreter.py Co-authored-by: Evan Pete Walsh <epwalsh10@gmail.com> * Update allennlp/interpret/influence_interpreters/influence_interpreter.py Co-authored-by: Evan Pete Walsh <epwalsh10@gmail.com> * Update allennlp/interpret/influence_interpreters/influence_interpreter.py Co-authored-by: Evan Pete Walsh <epwalsh10@gmail.com> * move some documentation of parameters to base class * delete one comment * delete one deprecated abstract method * changing interface * formatting * formatting err * passing mypy * passing mypy * passing mypy * passing mypy * passing integration test * passing integration test * adding a new option to the do-all function * modifying the callable function to the interface * update API, fixes * doc fixes * add `from_path` and `from_archive` methods * fix docs, improve logging * add test * address @matt-gardner's comments * fixes to documentation * update docs Co-authored-by: Evan Pete Walsh <epwalsh10@gmail.com> Co-authored-by: Evan Pete Walsh <petew@allenai.org>
dirkgr
added a commit
that referenced
this pull request
May 10, 2021
* Formatting * New activation functions * Makes position embeddings optional in the transformer embeddings * Adds T5 * Various fixes to make this start up * Share weights * Adds one test that passes, and one test that fails * use min_value_of_dtype in apply_mask * fixes, add beam search * encoder fixes * fix * fix beam search * fix tests * rename to just 'T5' * fix initialization from pretrained * add Model, DatasetReader, and Predictor * remove useless dataset reader * move high-level peices to allennlp-models * revert predictor changes * remove unneeded hidden_size * remove stray comment * bool masks * CHANGELOG * fix test file name * revert other change * revert other change * Distributed training with gradient accumulation (#5100) * Fixes distributed training with gradient accumulation * Fix in case we don't do anything in a batch group * Test for the problematic condition * Formatting * More formatting * Changelog * Fix another test * Fix even more tests * Fixes one more test * I can fix these tests all day. * Add link to gallery and demo in README (#5103) * Add link to gallery in README * Update README.md * try emojis Is this overkill? * Adding a metadata field to the basic classifier (#5104) * Adding metadata parameter to BasicClassifier * Fix * Updating the changelog * reformatting * updating parameter type * fixing import Co-authored-by: Dirk Groeneveld <dirkg@allenai.org> * additional W&B params (#5114) * additional W&B params * add wandb_kwargs * fix * fix docs * Add eval_mode argument to pretrained transformer embedder (#5111) * Add eval_mode argument to pretrained transformer embedder * Edit changelog entry * Lint * Update allennlp/modules/token_embedders/pretrained_transformer_embedder.py * Apply suggestions from code review Co-authored-by: Evan Pete Walsh <epwalsh10@gmail.com> Co-authored-by: Evan Pete Walsh <petew@allenai.org> * specify 'truncation' to avoid transformers warning (#5120) * specify 'truncation' to avoid transformers warning * Update docs * Remove `stride` param * Update CHANGELOG.md Co-authored-by: Dirk Groeneveld <dirkg@allenai.org> * Predicting with a dataset reader on a multitask model (#5115) * Create a way to use allennlp predict with a dataset and a multitask model * Fix type ignoration * Changelog * Fix to the predictor * fix bug with interleaving dataset reader (#5122) * fix bug with interleaving dataset reader * more tests * Update allennlp/data/dataset_readers/interleaving_dataset_reader.py * Update allennlp/data/dataset_readers/interleaving_dataset_reader.py * remove jsonpickle from dependencies (#5121) Co-authored-by: Dirk Groeneveld <dirkg@allenai.org> * Update docstring for basic_classifier (#5124) * improve error message from Registrable class (#5125) Co-authored-by: Akshita Bhagia <akshita23bhagia@gmail.com> * Prepare for release v2.3.0 * fix docs CI * Take the number of runs in the test for distributed metrics (#5127) * Take the number of runs in the test for distributed metrics * Changelog * Add influence functions to interpret module (#4988) * creating a new functionality to fields and instances to support outputing instnaces to json files * creating tests for the new functionality * fixing docs * Delete __init__.py * Delete influence_interpreter.py * Delete use_if.py * Delete simple_influence_test.py * fixing docs * finishing up SimpleInfluence * passing lint * passing format * making small progress in coding * Delete fast_influence.py Submit to the wrong branch * Delete faiss_utils.py wrong branch * Delete gpt2_bug.py not sure why it's included * Delete text_class.py not sure why it's included * adding test file * adding testing files * deleted unwanted files * deleted unwanted files and rearrange test files * small bug * adjust function call to save instance in json * Update allennlp/interpret/influence_interpreters/influence_interpreter.py Co-authored-by: Evan Pete Walsh <epwalsh10@gmail.com> * Update allennlp/interpret/influence_interpreters/influence_interpreter.py Co-authored-by: Evan Pete Walsh <epwalsh10@gmail.com> * Update allennlp/interpret/influence_interpreters/influence_interpreter.py Co-authored-by: Evan Pete Walsh <epwalsh10@gmail.com> * move some documentation of parameters to base class * delete one comment * delete one deprecated abstract method * changing interface * formatting * formatting err * passing mypy * passing mypy * passing mypy * passing mypy * passing integration test * passing integration test * adding a new option to the do-all function * modifying the callable function to the interface * update API, fixes * doc fixes * add `from_path` and `from_archive` methods * fix docs, improve logging * add test * address @matt-gardner's comments * fixes to documentation * update docs Co-authored-by: Evan Pete Walsh <epwalsh10@gmail.com> Co-authored-by: Evan Pete Walsh <petew@allenai.org> * Update CONTRIBUTING.md (#5133) * Update CONTRIBUTING.md * updated changelog Co-authored-by: Akshita Bhagia <akshita23bhagia@gmail.com> Co-authored-by: Arjun Subramonian <arjuns@ip-192-168-0-106.us-west-2.compute.internal> * fix #5132 (#5134) * fix * Prepare for release v2.3.1 * Fairness Metrics (#5093) * Added three definitions of fairness * Updated CHANGELOG * Added DemographicParityWithoutGroundTruth and finished tests * finished refactoring Independence, Separation, and Sufficiency to accumulate * added distributed functionality to Independence, Sufficiency, and Separation * Finished aggregate and distributed functionality for DemographicParityWithoutGroundTruth * fixed GPU and doc issues * fixed GPU and doc issues * fixed GPU and doc issues * fixed GPU issues * fixed GPU issues * added init file * fixed typo * minor docstring changes * minor changes to docstring * Added simple explanations of fairness metrics to docstrings * Further vectorized all metric implementations * Fixed device issue Co-authored-by: Arjun Subramonian <arjuns@Arjuns-MacBook-Pro.local> Co-authored-by: Akshita Bhagia <akshita23bhagia@gmail.com> Co-authored-by: Dirk Groeneveld <dirkg@allenai.org> * fix cached_path for hub downloads (#5141) * fix cached_path for hub downloads * fix test name * fix type hint * Update allennlp/common/file_utils.py Co-authored-by: Lysandre Debut <lysandre@huggingface.co> Co-authored-by: Lysandre Debut <lysandre@huggingface.co> * fix * fix Co-authored-by: epwalsh <epwalsh10@gmail.com> Co-authored-by: Evan Pete Walsh <petew@allenai.org> Co-authored-by: Jacob Morrison <jacob1morrison@gmail.com> Co-authored-by: Nelson Liu <nelson-liu@users.noreply.github.com> Co-authored-by: Akshita Bhagia <akshita23bhagia@gmail.com> Co-authored-by: Leo Liu <zeyuliu2@uw.edu> Co-authored-by: ArjunSubramonian <arjun.subramonian@gmail.com> Co-authored-by: Arjun Subramonian <arjuns@ip-192-168-0-106.us-west-2.compute.internal> Co-authored-by: Arjun Subramonian <arjuns@Arjuns-MacBook-Pro.local> Co-authored-by: Lysandre Debut <lysandre@huggingface.co>
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
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.
Feature enrichment.
Changes proposed in this pull request:
SimpleInfluence. This function will score training examples for each test example; the scores are indicators of how the training example affects the model's (learning and thus) predictions on the test example.