Issue 2153/ Refactor PyTorch 64-Bit Data Type Conversion #2153

[teelrabbit]

Made changes so it uses a function dtype_to_32bit that converts 64-bit data types to 32-bit data types.

[YifanShenSZ]

Thanks @teelrabbit for the proposed fix. We will need to also fix where all those "number to dtype" mapping are used, e.g. ops.py

[teelrabbit]

Thanks @teelrabbit for the proposed fix. We will need to also fix where all those "number to dtype" mapping are used, e.g. ops.py

Makes sense. I'll do some testing and see I can use the current solution across all instances of where the mapping is used 🤙🏻

[teelrabbit]

"number to dtype" mapping are used

Made some changes to the occurrences of NUM_TO_TORCH_DTYPE to use "dtype_to_32bit". Can you comfirm if this is what you intended by "number to dtype". https://pastes.dev/2oAOqJeDM1 @YifanShenSZ

db22515

'        np_type = nptype_from_builtin(target_dtype.dtype)
        dtype = NUMPY_DTYPE_TO_TORCH_NUM[np_type]	        dtype = NUMPY_DTYPE_TO_TORCH_NUM[np_type]


    torch_dtype = NUM_TO_TORCH_DTYPE[dtype]	    torch_dtype = dtype_to_32bit(dtype)
    if isinstance(_input, Var) and _input.can_be_folded_to_const():	    if isinstance(_input, Var) and _input.can_be_folded_to_const():
        # numpy -> torch -> torch cast -> numpy	        # numpy -> torch -> torch cast -> numpy
        # This path is needed to use the mapping of passed in dtypes to torch dtypes.	        # This path is needed to use the mapping of passed in dtypes to torch dtypes.

[YifanShenSZ]

One last minor revision needed, overall LGTM: As CI suggests, we should still keep the number keys 4 and 7, and update the corresponding values to 64-bit dtypes

Sorry about the review delay, we were on a change freeze due to 7.2 release 🙏 Many thanks for polishing coremltools!

Waiting for last minor revision

[YifanShenSZ]

LGTM! CI ✅ Many thanks for contributing to coremltools!

[teelrabbit]

LGTM! CI ✅ Many thanks for contributing to coremltools!

[teelrabbit]

Also this issue should be closed out #2153 (comment)