Add SegmentMax-16 reference implementation
#28788
Open
+1,132
−0
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
Signed-off-by: p-wysocki <[email protected]>
Signed-off-by: p-wysocki <[email protected]>
Signed-off-by: p-wysocki <[email protected]>
…into segmentmax_core Signed-off-by: p-wysocki <[email protected]>
Signed-off-by: p-wysocki <[email protected]>
Signed-off-by: p-wysocki <[email protected]>
Signed-off-by: p-wysocki <[email protected]>
Signed-off-by: p-wysocki <[email protected]>
Signed-off-by: p-wysocki <[email protected]>
Signed-off-by: p-wysocki <[email protected]>
Signed-off-by: p-wysocki <[email protected]>
Signed-off-by: p-wysocki <[email protected]>
Signed-off-by: p-wysocki <[email protected]>
Signed-off-by: p-wysocki <[email protected]>
Signed-off-by: p-wysocki <[email protected]>
Signed-off-by: p-wysocki <[email protected]>
Signed-off-by: p-wysocki <[email protected]>
…into segmentmax_core Signed-off-by: p-wysocki <[email protected]>
Signed-off-by: p-wysocki <[email protected]>
Signed-off-by: p-wysocki <[email protected]>
Signed-off-by: p-wysocki <[email protected]>
Signed-off-by: p-wysocki <[email protected]>
Signed-off-by: p-wysocki <[email protected]>
github-actions
bot
added
category: Core
This was referenced Feb 3, 2025
t-jankowski
reviewed
Feb 4, 2025
| @@ -20,6 +20,12 @@ enum class PadMode { CONSTANT = 0, EDGE, REFLECT, SYMMETRIC }; | |||
| OPENVINO_API | |||
| std::ostream& operator<<(std::ostream& s, const PadMode& type); | |||
|
|
|||
| /// \brief Fill modes for the `SegmentMax` operator. | |||
There was a problem hiding this comment.
If this is specific for SegmentMax op only, then it should be placed in the op dedicated namespace and file.
Otherwise generalize the comment.
There was a problem hiding this comment.
There is a chance to extend it and reuse it with other ops in the future like SegmentMin
Suggested change
| /// \brief Fill modes for the `SegmentMax` operator. | |
| /// \brief Fill modes to set default value for operators like `SegmentMax`. |
But if it's preferred to use a separate enums for such ops, it should be inside the SegmentMax class/namespace/file.
praasz
reviewed
Feb 4, 2025
Comment on lines
+24
to
+25
| const size_t inner_dim_size = | ||
| std::accumulate(data_shape.begin() + 1, data_shape.end(), 1, std::multiplies<size_t>()); |
There was a problem hiding this comment.
Suggested change
| const size_t inner_dim_size = | |
| std::accumulate(data_shape.begin() + 1, data_shape.end(), 1, std::multiplies<size_t>()); | |
| const auto inner_dim_size = shape_size(data_shape.begin() +1, data_shape.end()); |
Comment on lines
+26
to
+27
| std::vector<std::vector<T>> max_values(num_segments, | ||
| std::vector<T>(inner_dim_size, std::numeric_limits<T>::lowest())); |
There was a problem hiding this comment.
Can be input data copy avoided and set max values in the result tensor instead?
mitruska
reviewed
Feb 5, 2025
Comment on lines
+16
to
+19
| template <typename T, typename T_idx> | ||
| void segment_max(const T* data, | ||
| const Shape& data_shape, | ||
| const T_idx* segment_ids, |
There was a problem hiding this comment.
To avoid additional type specialization for int32 and int64 (in favour of smaller binary size) there is a common approach to upcast the i32 to i64, and use only i64 in the ref instead of T_idx.
But as this reference implementation is not used in the evaluate from the op class, it's not crucial.
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.
Details:
SegmentMax-16to Core #28698, it's currently merged into this PRRelated PRs:
SegmentMax-16#28103SegmentMax-16to Core #28698Tickets: