Fix float16 trunc division rounding

[Cpthimself (Alphaxiaoteng)]

Summary

Fixes #990.

ONNX Runtime can evaluate the float16 division in aten::div(..., rounding_mode="trunc") with a higher-precision intermediate than PyTorch. Applying truncation to that value can produce an off-by-one result at float16 boundaries.

Materialize the float16 quotient before truncation, matching PyTorch's float16 rounding semantics. The existing float16 trunc_rounding regression case is enabled again.

Tests

• python -m pytest tests/function_libs/torch_lib/ops_test.py -k test_output_match_opinfo__div_mode -vv
• ruff check onnxscript/function_libs/torch_lib/ops/core.py tests/function_libs/torch_lib/ops_test_data.py
• ruff format --check onnxscript/function_libs/torch_lib/ops/core.py tests/function_libs/torch_lib/ops_test_data.py
• git diff --check

[Copilot]

Pull request overview

This PR fixes the float16 off-by-one mismatch for aten::div(..., rounding_mode="trunc") by ensuring the division result is rounded to float16 before applying truncation, matching PyTorch’s float16 semantics and re-enabling the corresponding regression coverage (issue #990).

Changes:

• In aten_div_mode(..., rounding_mode="trunc"), materialize the float16 quotient via an explicit cast before calling aten_trunc.
• Re-enable the float16 trunc_rounding variant in the TorchLib OpInfo test data by removing the skip for issue #990.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
onnxscript/function_libs/torch_lib/ops/core.py	Casts the float16 division quotient to float16 before truncation to match PyTorch rounding semantics.
tests/function_libs/torch_lib/ops_test_data.py	Removes the float16 skip for the div_mode trunc-rounding variant to restore regression coverage.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 72.64%. Comparing base (3ba2bf7) to head (f34edf1).
⚠️ Report is 3 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #3025   +/-   ##
=======================================
  Coverage   72.64%   72.64%           
=======================================
  Files         265      265           
  Lines       32251    32254    +3     
  Branches     3050     3051    +1     
=======================================
+ Hits        23429    23432    +3     
  Misses       7786     7786           
  Partials     1036     1036           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

[Cpthimself (Alphaxiaoteng)]

@microsoft-github-policy-service agree

[Justin Chu (justinchuby)]

Thanks!