Closed yz-tang closed 4 months ago
When I was using flashinfer, I encountered that the heads of some models were not powers of 2. I refer to flashinfer/python/tests/alibi_reference.py, modifies this part of the C++ code.
When I was using flashinfer, I encountered that the heads of some models were not powers of 2. I refer to flashinfer/python/tests/alibi_reference.py, modifies this part of the C++ code.