Update metrics - #321
Merged
Merged
Conversation
for more information, see https://pre-commit.ci
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Code Review
This pull request refactors several evaluation metrics (including MRR, HitRate, and NegativeLogLikelihood) to support averaging on a per-trip basis instead of per-batch, introduces a new MeanRank metric, and updates the basket dataset to calculate sample counts more efficiently. The review feedback highlights critical TensorFlow type mismatch issues in the MRR and MeanRank metrics that could lead to runtime errors, and suggests a more Pythonic implementation for the new get_n_samples method.
Comment on lines
+171
to
+173
| else: | ||
| self.mrr.assign(self.mrr + tf.reduce_sum(mean_rank)) | ||
| self.n_evals.assign(self.n_evals + tf.cast(tf.shape(y_true)[0], tf.float32)) | ||
| self.mrr.assign(self.mrr + tf.reduce_sum(tf.cast(1 / item_ranks, dtype=tf.float32))) | ||
| self.n_evals.assign(self.n_evals + tf.shape(y_true)[0]) |
Contributor
There was a problem hiding this comment.
Using tf.shape(y_true)[0] directly without casting will raise a TypeError in TensorFlow due to type mismatch between the float32 variable self.n_evals and the int32/int64 tensor returned by tf.shape. Additionally, using assign_add is more efficient and standard for updating metric variables.
Suggested change
| else: | |
| self.mrr.assign(self.mrr + tf.reduce_sum(mean_rank)) | |
| self.n_evals.assign(self.n_evals + tf.cast(tf.shape(y_true)[0], tf.float32)) | |
| self.mrr.assign(self.mrr + tf.reduce_sum(tf.cast(1 / item_ranks, dtype=tf.float32))) | |
| self.n_evals.assign(self.n_evals + tf.shape(y_true)[0]) | |
| else: | |
| self.mrr.assign_add(tf.reduce_sum(tf.cast(1 / item_ranks, dtype=tf.float32))) | |
| self.n_evals.assign_add(tf.cast(tf.shape(y_true)[0], self.n_evals.dtype)) |
Comment on lines
+231
to
+237
| if batch is not None and self.average_on_trip: | ||
| int_batch = tf.cast(batch, tf.int32) | ||
| self.mr.assign(self.mr + tf.reduce_sum(tf.math.segment_mean(float_rank, int_batch))) | ||
| self.n_evals.assign(self.n_evals + tf.reduce_max(batch) + 1) | ||
| else: | ||
| self.mr.assign(self.mr + tf.reduce_sum(float_rank)) | ||
| self.n_evals.assign(self.n_evals + tf.cast(tf.shape(y_true)[0], tf.float32)) |
Contributor
There was a problem hiding this comment.
The current implementation of MeanRank under average_on_trip has several issues:
tf.math.segment_meanrequires sorted segment IDs, which is not guaranteed for arbitrary batches.tf.reduce_max(batch) + 1assumes batch indices are 0-indexed and contiguous, which is fragile and prone to bugs.- Adding
tf.reduce_max(batch)(int32) directly toself.n_evals(float32) will raise aTypeErrorin TensorFlow.
Using tf.unique and tf.math.unsorted_segment_mean with assign_add is much more robust, consistent with other metrics in this file, and avoids type mismatch errors.
if batch is not None and self.average_on_trip:
unique_trips, segment_ids = tf.unique(batch)
trip_means = tf.math.unsorted_segment_mean(
float_rank, segment_ids, tf.shape(unique_trips)[0]
)
self.mr.assign_add(tf.reduce_sum(trip_means))
self.n_evals.assign_add(tf.cast(tf.shape(unique_trips)[0], self.n_evals.dtype))
else:
self.mr.assign_add(tf.reduce_sum(float_rank))
self.n_evals.assign_add(tf.cast(tf.shape(y_true)[0], self.n_evals.dtype))
Comment on lines
+207
to
+213
| def get_n_samples(self) -> int: | ||
| """Return the numbers of samples/transations of the TripDataset.""" | ||
| n_samples = 0 | ||
| for trip in self.trips: | ||
| n_samples += len(trip.purchases) | ||
|
|
||
| return n_samples |
Contributor
There was a problem hiding this comment.
The get_n_samples method can be written more pythonically and efficiently using a generator expression with sum(). Also, corrected typos in the docstring ('numbers' -> 'number', 'transations' -> 'transactions').
Suggested change
| def get_n_samples(self) -> int: | |
| """Return the numbers of samples/transations of the TripDataset.""" | |
| n_samples = 0 | |
| for trip in self.trips: | |
| n_samples += len(trip.purchases) | |
| return n_samples | |
| def get_n_samples(self) -> int: | |
| """Return the number of samples/transactions of the TripDataset.""" | |
| return sum(len(trip.purchases) for trip in self.trips) |
Contributor
Contributor
Coverage Report for Python 3.12
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Contributor
Coverage Report for Python 3.9
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Contributor
Coverage Report for Python 3.11
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
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
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.
Description of the goal of the PR
Description:
Changes this PR introduces (fill it before implementation)
Checklist before requesting a review