TF 2.2 documentation update

[mchoi8739]

Description of changes:

Update TF documentation, pointing out the newest feature with that latest Debugger pySDK and smdebug release.

Style and formatting:

I have run pre-commit install to ensure that auto-formatting happens with every commit.

Issue number, if available

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[codecov-commenter]

Codecov Report

Merging #322 into master will decrease coverage by 21.14%. The diff coverage is n/a.

@@             Coverage Diff             @@
##           master     #322       +/-   ##
===========================================
- Coverage   85.29%   64.15%   -21.15%     
===========================================
  Files          86       86               
  Lines        6352     6352               
===========================================
- Hits         5418     4075     -1343     
- Misses        934     2277     +1343     

Impacted Files	Coverage Δ
smdebug/tensorflow/__init__.py	0.00% <0.00%> (-100.00%)	:arrow_down:
smdebug/tensorflow/constants.py	0.00% <0.00%> (-100.00%)	:arrow_down:
smdebug/tensorflow/singleton_utils.py	0.00% <0.00%> (-100.00%)	:arrow_down:
smdebug/tensorflow/collection.py	0.00% <0.00%> (-95.88%)	:arrow_down:
smdebug/tensorflow/keras.py	0.00% <0.00%> (-93.59%)	:arrow_down:
smdebug/tensorflow/session.py	0.00% <0.00%> (-91.87%)	:arrow_down:
smdebug/tensorflow/utils.py	0.00% <0.00%> (-87.77%)	:arrow_down:
smdebug/tensorflow/callable_cache.py	0.00% <0.00%> (-82.61%)	:arrow_down:
smdebug/tensorflow/base_hook.py	0.00% <0.00%> (-79.01%)	:arrow_down:
smdebug/tensorflow/tensor_ref.py	11.29% <0.00%> (-77.42%)	:arrow_down:
... and 24 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update e38f381...1eee9c6. Read the comment docs.

[Vikas-kum]

On main doc page there is a Note: Limited support of the zero script change experience for TensorFlow 2.2. The tensor collections layers, inputs, outputs, and gradients are currently not available

I think you would need to remove that.

[mchoi8739]

On main doc page there is a Note: Limited support of the zero script change experience for TensorFlow 2.2. The tensor collections layers, inputs, outputs, and gradients are currently not available

I think you would need to remove that.

I will do after I verify by running the test notebook that I found the image issue with cuda version. I want to make sure if everything working is true for customers.

[NihalHarish]

Can we also add a section on the version and behavior of smdebug shipped on different DLCs

[mchoi8739]

Confirm that the bug has fixed. Removing the limitation note.

[mchoi8739]

Can we also add a section on the version and behavior of smdebug shipped on different DLCs

I am not sure what are the different behaviors would be? I thought it works in the same way for all Debugger-supported DLCs. Can you tell me what are those behaviors?

[Vikas-kum]

On main doc page there is a Note: Limited support of the zero script change experience for TensorFlow 2.2. The tensor collections layers, inputs, outputs, and gradients are currently not available I think you would need to remove that.

I will do after I verify by running the test notebook that I found the image issue with cuda version. I want to make sure if everything working is true for customers.

"want to make sure if everything working is true for customers." -> This should be done before the docs merged in and made public. I hope it is working now.

[mchoi8739]

On main doc page there is a Note: Limited support of the zero script change experience for TensorFlow 2.2. The tensor collections layers, inputs, outputs, and gradients are currently not available I think you would need to remove that.

I will do after I verify by running the test notebook that I found the image issue with cuda version. I want to make sure if everything working is true for customers.

"want to make sure if everything working is true for customers." -> This should be done before the docs merged in and made public. I hope it is working now.

yes, I confirmed that it works now. I tested with an example notebook.

[NihalHarish]

Can we also add a section on the version and behavior of smdebug shipped on different DLCs

I am not sure what are the different behavior would be. I thought it works in the same way for all Debugger-supported DLCs. Can you tell me what are those behaviors?

In the case of TF 2.x containers, inputs, outputs, gradients, and intermediate layer are only saved for the TF 2.2.0 container

[mchoi8739]

Can we also add a section on the version and behavior of smdebug shipped on different DLCs

I am not sure what are the different behavior would be. I thought it works in the same way for all Debugger-supported DLCs. Can you tell me what are those behaviors?

In the case of TF 2.x containers, inputs, outputs, gradients, and intermediate layer are only saved for the TF 2.2.0 container

ok, updating the note that I put here: https://github.com/mchoi8739/sagemaker-debugger/blob/master/docs/tensorflow.md#debugger-built-in-tensor-collections-for-tensorflow