Fitting a multi output GP to different data sets repeatedly returns the following error in line 127:
qU_var_r = tdot(self.qU_var_r_W) + np.diag(self.qU_var_r_diag)
ValueError: operands could not be broadcast together with shapes (4,4) (10,10)
The shape of tdot(self.qU_var_r_W) scales with the number of output dimensions in Y (in this case 4).
The shape of np.diag(self.qU_var_r_diag) remains constant at (10, 10).
Thus the only Y training data sets which proceed past line 127 are those with specifically 10 output dimensions. This appears to be an error.
Any advice would be greatly appreciated. I would be very happy to provide further details if someone may be able to look into this.
eymc
commented
3 years ago
Fitting a multi output GP to different data sets repeatedly returns the following error in line 127:
qU_var_r = tdot(self.qU_var_r_W) + np.diag(self.qU_var_r_diag) ValueError: operands could not be broadcast together with shapes (4,4) (10,10)
The shape of tdot(self.qU_var_r_W) scales with the number of output dimensions in Y (in this case 4). The shape of np.diag(self.qU_var_r_diag) remains constant at (10, 10). Thus the only Y training data sets which proceed past line 127 are those with specifically 10 output dimensions. This appears to be an error.
Any advice would be greatly appreciated. I would be very happy to provide further details if someone may be able to look into this.