Linux-cpp-lisp/opt_einsum_fx
Einsum optimization using opt_einsum and PyTorch FX graph rewriting
PythonMIT
Issues
- 2
- 1
tensordot Expected a value of type 'int' for argument 'dims' but instead found type 'List[int]'
#16 opened by mariogeiger - 0
- 0
[Possible feature] Allow fusing non-constant scalars through "linear" operations
#7 opened by Linux-cpp-lisp - 0
[Feature] Add more "linear" operations
#6 opened by Linux-cpp-lisp - 1
[idea] fuse and move mul/div
#2 opened by mariogeiger - 0
[Feature] convert transpose into einsum
#5 opened by mariogeiger - 0
- 0
- 1
[idea] fuse einsums
#1 opened by mariogeiger