Jchmura/edge feature quantization - #745

Draft

jchmura-sc wants to merge 83 commits into

mainfrom

jchmura/edge-feature-quantization

jchmura-sc commented Aug 12, 2026

Collaborator

Scope of work done

Where is the documentation for this feature?: N/A

Did you add automated tests or write a test plan?

Updated Changelog.md? NO

Ready for code review?: NO

jchmura-sc added 30 commits

August 3, 2026 20:42


          Compile new protos

97f938a


          Add quantization ops


          Minor format updates

3b9134e


          Add tests

17a9d84


          Whitespace

4376ea7

WIP

838f2ed


          Add preprocessor diff

d357421


          Add test

459b8cb


          Update read side

be34beb


          Add integration code

be4a9fe


          Add docs


          Update docs

ec88d0f


          Add inline comment explaining uint16 cast

3e76a52


          Add quantize dequantize roundtrip test

c0b4847


          Simplify test logic

7c70b02


          Defensive check against nan or inf

1510b43


          Pass clip args directly to quantize


          Merge branch 'jchmura/feature_quant_ops' into jchmura/feature_quant_p…

f670c83

…reprocessor

Upd

d74590c

Upd

1ebbd19

upd

1fd3e28

WIP

48bab63

WIP

1ecf397

WIP

2fd2c64


          Add SUPPORTED_QUANTIZATION_BITS const

3fbe66a


          Add docstring to quantization transform

109e1e5


          Merge main

02bb49a


          Format docs

7ed4b29


          Revert diff

b490889


          Revert diff in graph types

42676f5

jchmura-sc added 2 commits

August 12, 2026 20:20


          Fix heterogenous sampler collate with only partially quantized node t…

5de7dd9

…ypes


          Merge branch 'jchmura/feature_quant_integration' into jchmura/edge-fe…

eccc0ea

…ature-quantization

jchmura-sc self-assigned this


          Remove planning docs

a528164

jchmura-sc commented

View reviewed changes

gigl/distributed/dist_dataset.py

jchmura-sc commented

View reviewed changes

gigl/src/data_preprocessor/lib/transform/utils.py Outdated

jchmura-sc commented

View reviewed changes

gigl/distributed/base_sampler.py Outdated

jchmura-sc commented

View reviewed changes

gigl/distributed/dist_range_partitioner.py

jchmura-sc commented

View reviewed changes

gigl/src/data_preprocessor/lib/transform/feature_quantization.py Outdated

jchmura-sc commented

View reviewed changes

gigl/src/data_preprocessor/lib/transform/feature_quantization.py Outdated

jchmura-sc commented

View reviewed changes

gigl/src/data_preprocessor/lib/transform/feature_quantization.py

Comment on lines +166 to +169

+                  if packed_feature_key in batch.schema.names:
+                      raise ValueError(
+                          f"Reserved packed feature key {packed_feature_key} already exists in the logical schema."
+                      )

jchmura-sc Aug 13, 2026

Collaborator Author

This or the check at the beggining of the quantization can be removed, as they check the same thing

jchmura-sc commented

View reviewed changes

gigl/src/data_preprocessor/lib/transform/feature_quantization.py

jchmura-sc commented

View reviewed changes

gigl/distributed/utils/neighborloader.py

jchmura-sc commented

View reviewed changes

gigl/distributed/utils/neighborloader.py Outdated

jchmura-sc commented

View reviewed changes

gigl/common/data/load_torch_tensors.py

jchmura-sc added 14 commits

August 13, 2026 17:18


          Simple improvements

d141185


          Clarify sampling weight quantization metadata

7dfee9c


          Simplify sampling weight metadata adjustment

10c6574


          Inline sampled edge metadata mapping

828ef2d


          Format edge quantization changes

06c3b71


          Improved docs for materialize_quantized_node_features

20bd8b4


          Return partitioned quantized edge features

03b0e49


          Minimize range partitioner documentation diff

d0f62cd


          Simplify edge partitioner diffs

3e58bd9


          Revert diff

4bcab57


          Remove redundant diff

a488153


          Merge branch 'jchmura/feature_quant_integration' into jchmura/edge-fe…

df6517b

…ature-quantization


          Merge main

df98e0f


          Extract common helper function for scatter quantized feats

34ea459

Base automatically changed from jchmura/feature_quant_integration to main

August 13, 2026 21:02

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

Reviewers

At least 2 approving reviews are required to merge this pull request.

Labels

None yet