1 problem
- 0 votes0 replies1 view
Weak identifiability of multi-head attention parameters
MHA weak identifiability conjecture. For every parameter of an MHA model, the parameter is weakly identifiable with respect to the symmetry group
Weak identifiability of multi-head attention parameters
MHA weak identifiability conjecture. For every parameter of an MHA model, the parameter is weakly identifiable with respect to the symmetry group