Skip to content

Adding Pipeline methods [WIP]#114

Open
khizirsiddiqui wants to merge 5 commits into
SforAiDl:masterfrom
khizirsiddiqui:add-pipeline
Open

Adding Pipeline methods [WIP]#114
khizirsiddiqui wants to merge 5 commits into
SforAiDl:masterfrom
khizirsiddiqui:add-pipeline

Conversation

@khizirsiddiqui
Copy link
Copy Markdown
Collaborator

No description provided.

Comment thread KD_Lib/utils/pipeline.py
methods.

:param: steps (list) list of KD_Lib.KD or KD_Lib.Pruning or KD_Lib.Quantization
:param: epochs (int) number of iterations through whole batch for each method in
Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We can remove this parameter right? Since each individual method will have epochs in its initialization

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Distillation specifies these parameters in train_student method instead of class constructor.

Comment thread KD_Lib/utils/pipeline.py Outdated
Comment on lines +74 to +81
if hasattr(process, 'train_student'):
process.train_student(self.epochs, self.plot_losses, self.save_model, self.save_model_path)
elif hasattr(proces, 'prune'):
process.prune()
elif hasattr(process, 'quantize'):
process.quantize()
else:
raise TypeError("{} is not supported by the pipeline yet."
Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Save the final model after all the steps in the process

@codecov-commenter
Copy link
Copy Markdown

codecov-commenter commented Jun 11, 2021

Codecov Report

Merging #114 (8f8d42b) into master (0ecccec) will decrease coverage by 55.04%.
The diff coverage is 8.92%.

@@            Coverage Diff             @@
##           master    #114       +/-   ##
==========================================
- Coverage   55.83%   0.79%   -55.05%     
==========================================
  Files          60      61        +1     
  Lines        2115    2012      -103     
==========================================
- Hits         1181      16     -1165     
- Misses        934    1996     +1062     
Impacted Files Coverage Δ
KD_Lib/utils/pipeline.py 7.27% <7.27%> (ø)
KD_Lib/utils/__init__.py 100.00% <100.00%> (ø)
KD_Lib/models/nin.py 0.00% <0.00%> (-100.00%) ⬇️
KD_Lib/models/lstm.py 0.00% <0.00%> (-100.00%) ⬇️
KD_Lib/models/lenet.py 0.00% <0.00%> (-100.00%) ⬇️
KD_Lib/models/shallow.py 0.00% <0.00%> (-100.00%) ⬇️
KD_Lib/models/__init__.py 0.00% <0.00%> (-100.00%) ⬇️
KD_Lib/KD/text/__init__.py 0.00% <0.00%> (-100.00%) ⬇️
KD_Lib/Pruning/__init__.py 0.00% <0.00%> (-100.00%) ⬇️
KD_Lib/KD/vision/KA/__init__.py 0.00% <0.00%> (-100.00%) ⬇️
... and 50 more

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Making a pipeline for Pruning, Quantization and Knowledge Distillation

3 participants