[GitHub] ptrendx opened a new pull request #14173: [WIP] MXNet AMP (automatic mixed precision)

[GitBox <gi...@apache.org>]

ptrendx opened a new pull request #14173: [WIP] MXNet AMP (automatic mixed precision)
URL: https://github.com/apache/incubator-mxnet/pull/14173
 
 
   ## Description ##
   
   Whis is a Work in Progress PR for AMP (automatic mixed precision) support for MXNet, similar to pyTorch version found in https://github.com/NVIDIA/apex.
   
   This PR relies on multiple other PRs and bug fixes, listed in Comments section.
   
   Dynamic loss scaling part done by @Caenorst (commits were squashed for easier rebasing).
   
   ## Checklist ##
   ### Essentials ###
   Please feel free to remove inapplicable items for your PR.
   - [ ] Changes are complete (i.e. I finished coding on this PR)
   - [ ] All changes have test coverage:
   - Unit tests are added for small changes to verify correctness (e.g. adding a new operator)
   - Nightly tests are added for complicated/long-running ones (e.g. changing distributed kvstore)
   - [ ] Code is well-documented: 
   - For user-facing API changes, API doc string has been updated. 
   - For new C++ functions in header files, their functionalities and arguments are documented. 
   - For new examples, README.md is added to explain the what the example does, the source of the dataset, expected performance on test set and reference to the original paper if applicable
   - Check the API doc at http://mxnet-ci-doc.s3-accelerate.dualstack.amazonaws.com/PR-$PR_ID/$BUILD_ID/index.html
   - [ ] To the my best knowledge, examples are either not affected by this change, or have been fixed to be compatible with this change
   
   ### Changes ###
   - [x] Auditor that enables/disables operations to be done in FP16 automatically. It is implemented via patching MXNet functions in mxnet.symbol and mxnet.ndarray to insert casts to FP16/FP32 where necessary.
   - [x] Operator `amp_cast` and `amp_multicast` that handle casting between FP16/FP32 when necessary and do not change other types. They are optimized to not do anything if the input is already in the proper type.
   - [ ] Dynamic loss scaling and supporting operators for checking gradients for infs/NaNs and skipping update step if such value is encountered.
   
   ## Comments ##
   - This PR relies on multiple other PRs/bug fixes:
     - [ ] #14153 
     - [ ] https://github.com/dmlc/tvm/pull/2572 - once this is done, need to change submodule to point to dmlc/tvm again
     - [x] #14097 
     - [ ] https://github.com/dmlc/dmlc-core/issues/503
     - [ ] #12139 which was masked by #12189, but which prevents updating to newer dmlc-core
   - Once 
   
   FYI @eric-haibin-lin @szha 
   

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services