MIGraphX abstraction layer for torch converters

Open shivadbhavsar opened this issue 2 years ago • 1 comments

Figure out specs and high level design for this functionallity. Goal: Avoid adding ops directly in torch_migraphx converters incase op specs change within migraphx.

Nov 10 '23 18:11 shivadbhavsar

How exactly would this be better than the current method?

Jun 03 '24 21:06 CharlieL7