Dummy forward#953
Merged
mzr1996 merged 1 commit intoopen-mmlab:masterfrom Aug 8, 2022
twmht:dummy_forward
Merged
Conversation
Codecov Report
@@ Coverage Diff @@
## master #953 +/- ##
==========================================
+ Coverage 84.57% 85.43% +0.86%
==========================================
Files 134 134
Lines 8772 8777 +5
Branches 1516 1516
==========================================
+ Hits 7419 7499 +80
+ Misses 1117 1056 -61
+ Partials 236 222 -14
Flags with carried forward coverage won't be shown. Click here to find out more.
Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here. |
Contributor
Author
|
can you help me to check CI jobs? |
mzr1996
reviewed
Aug 8, 2022
mmcls/models/classifiers/image.py
Outdated
| See `mmclassificaiton/tools/analysis_tools/get_flops.py` | ||
| """ | ||
| x = self.extract_feat(img) | ||
| outs = self.head.simple_test(x, post_process=False) |
Member
There was a problem hiding this comment.
Should the forward_dummy includes head.simple_test?
For the classification tasks, we may use different datasets to do pre-training. If the forward_dummy includes head.simple_test, the flops value of the same model trained on ImageNet-1k and ImageNet-21k will be different. It's a little confusing.
mzr1996
approved these changes
Aug 8, 2022
Member
|
The unit tests need to be fixed. |
Contributor
Author
sure. |
mzr1996
pushed a commit
that referenced
this pull request
Aug 22, 2022
Ezra-Yu
pushed a commit
to Ezra-Yu/mmclassification
that referenced
this pull request
Sep 6, 2022
Ezra-Yu
pushed a commit
to Ezra-Yu/mmclassification
that referenced
this pull request
Sep 6, 2022
* [Improve] Use `forward_dummy` to calculate FLOPS. (open-mmlab#953) * fixed Co-authored-by: Ming-Hsuan-Tu <alec.tu@acer.com>
mzr1996
added a commit
that referenced
this pull request
Dec 30, 2022
* unit test for multi_task_head * [Feature] MultiTaskHead (#628, #481) * [Fix] lint for multi_task_head * [Feature] Add `MultiTaskDataset` to support multi-task training. * Update MultiTaskClsHead * Update docs * [CI] Add test mim CI. (#879) * [Fix] Remove duplicated wide-resnet metafile. * [Feature] Support MPS device. (#894) * [Feature] Support MPS device. * Add `auto_select_device` * Add unit tests * [Fix] Fix Albu crash bug. (#918) * Fix albu BUG: using albu will cause the label from array(x) to array([x]) and crash the trainning * Fix common * Using copy incase potential bug in multi-label tasks * Improve coding * Improve code logic * Add unit test * Fix typo * Fix yapf * Bump version to 0.23.2. (#937) * [Improve] Use `forward_dummy` to calculate FLOPS. (#953) * Update README * [Docs] Fix typo for wrong reference. (#1036) * [Doc] Fix typo in tutorial 2 (#1043) * [Docs] Fix a typo in ImageClassifier (#1050) * add mask to loss * add another pipeline * adpat the pipeline if there is no mask * switch mask and task * first version of multi data smaple * fix problem with attribut by getattr * rm img_label suffix, fix 'LabelData' object has no attribute 'gt_label' * training without evaluation * first version work * add others metrics * delete evaluation from dataset * fix linter * fix linter * multi metrics * first version of test * change evaluate metric * Update tests/test_models/test_heads.py Co-authored-by: Colle <piercus@users.noreply.github.com> * Update tests/test_models/test_heads.py Co-authored-by: Colle <piercus@users.noreply.github.com> * add tests * add test for multidatasample * create a generic test * create a generic test * create a generic test * change multi data sample * correct test * test * add new test * add test for dataset * correct test * correct test * correct test * correct test * fix : #5 * run yapf * fix linter * fix linter * fix linter * fix isort * fix isort * fix docformmater * fix docformmater * fix linter * fix linter * fix data sample * Update mmcls/structures/multi_task_data_sample.py Co-authored-by: Colle <piercus@users.noreply.github.com> * Update mmcls/structures/multi_task_data_sample.py Co-authored-by: Colle <piercus@users.noreply.github.com> * Update mmcls/structures/multi_task_data_sample.py Co-authored-by: Colle <piercus@users.noreply.github.com> * Update mmcls/structures/multi_task_data_sample.py Co-authored-by: Colle <piercus@users.noreply.github.com> * Update mmcls/structures/multi_task_data_sample.py Co-authored-by: Colle <piercus@users.noreply.github.com> * Update mmcls/structures/multi_task_data_sample.py Co-authored-by: Colle <piercus@users.noreply.github.com> * Update tests/test_structures/test_datasample.py Co-authored-by: Colle <piercus@users.noreply.github.com> * Update mmcls/structures/multi_task_data_sample.py Co-authored-by: Colle <piercus@users.noreply.github.com> * Update tests/test_structures/test_datasample.py Co-authored-by: Colle <piercus@users.noreply.github.com> * Update tests/test_structures/test_datasample.py Co-authored-by: Colle <piercus@users.noreply.github.com> * update data sample * update head * update head * update multi data sample * fix linter * fix linter * fix linter * fix linter * fix linter * fix linter * update head * fix problem we don't set pred or gt * fix problem we don't set pred or gt * fix problem we don't set pred or gt * fix linter * fix : #2 * fix : linter * update multi head * fix linter * fix linter * update data sample * update data sample * fix ; linter * update test * test pipeline * update pipeline * update test * update dataset * update dataset * fix linter * fix linter * update formatting * add test for multi-task-eval * update formatting * fix linter * update test * update * add test * update metrics * update metrics * add doc for functions * fix linter * training for multitask 1.x * fix linter * run flake8 * run linter * update test * add mask in evaluation * update metric doc * update metric doc * Update mmcls/evaluation/metrics/multi_task.py Co-authored-by: Colle <piercus@users.noreply.github.com> * Update mmcls/evaluation/metrics/multi_task.py Co-authored-by: Colle <piercus@users.noreply.github.com> * Update mmcls/evaluation/metrics/multi_task.py Co-authored-by: Colle <piercus@users.noreply.github.com> * Update mmcls/evaluation/metrics/multi_task.py Co-authored-by: Colle <piercus@users.noreply.github.com> * Update mmcls/evaluation/metrics/multi_task.py Co-authored-by: Colle <piercus@users.noreply.github.com> * Update mmcls/evaluation/metrics/multi_task.py Co-authored-by: Colle <piercus@users.noreply.github.com> * update metric doc * update metric doc * Fix cannot import name MultiTaskDataSample * fix test_datasets * fix test_datasets * fix linter * add an example of multitask * change name of configs dataset * Refactor the multi-task support * correct test and metric * add test to multidatasample * add test to multidatasample * correct test * correct metrics and clshead * Update mmcls/models/heads/cls_head.py Co-authored-by: Colle <piercus@users.noreply.github.com> * update cls_head.py documentation * lint * lint * fix: lint * fix linter * add eval mask * fix documentation * fix: single_label.py back to 1.x * Update mmcls/models/heads/multi_task_head.py Co-authored-by: Ma Zerun <mzr1996@163.com> * Remove multi-task configs. Co-authored-by: mzr1996 <mzr1996@163.com> Co-authored-by: HinGwenWoong <peterhuang0323@qq.com> Co-authored-by: Ming-Hsuan-Tu <alec.tu@acer.com> Co-authored-by: Lei Lei <18294546+Crescent-Saturn@users.noreply.github.com> Co-authored-by: WRH <12756472+wangruohui@users.noreply.github.com> Co-authored-by: marouaneamz <maroineamil99@gmail.com> Co-authored-by: marouane amzil <53240092+marouaneamz@users.noreply.github.com>
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
2 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.
We should add
forward_dummyto query flops by mmcv (https://github.com/open-mmlab/mmcv/blob/master/mmcv/cnn/utils/flops_counter.py#L38). Althugh we already haveextract_feat, but it's not sync with mmdetection(https://github.com/open-mmlab/mmdetection/blob/master/mmdet/models/detectors/single_stage.py#L48).we should sync both projects since many people may want to put both together into a single project.