Has it been any update or progress in this line? I recently encountered a similar problem that I posted in Nested and different AD methods altogether: How to add AD calculations inside my loss function when using neural differential equations? that I am trying to make work. Thanks!
1 Like