onnxruntime
Fix memory consumption discrepancy
#12266
Merged

Commits
  • release cached cuda memory after temp model_copy run
    pengwa committed 3 years ago
  • op schema change only: remove PythonOp forward output from PythonOpGrad inputs.
    pengwa committed 3 years ago
  • always export model using torch.no_grad
    pengwa committed 3 years ago
  • 1.update PythonOP's "input_requires_grads" attribute according to ORT gradient graph.
    pengwa committed 3 years ago
  • fix uts
    pengwa committed 3 years ago
  • refine basde on wschin's comments && fix pylint
    pengwa committed 3 years ago
  • fix comments
    pengwa committed 3 years ago
  • fix unused variable
    pengwa committed 3 years ago
Loading