3cb22adccb2deed60e56cd46491b26ed4209d6f4,qucumber/quantum_reconstruction.py,QuantumReconstruction,compute_batch_gradients,#QuantumReconstruction#Any#Any#Any#,50

Before Change

        // Positive Phase
        for i in range(samples_batch.shape[0]):
            if bases_batch is None:
                for par in rbm.state_dict():
                    grad["rbm_am"][par] += self.nn_state.gradient(samples_batch[i])["rbm_am"][par]/float(samples_batch.shape[0])
            else:
                b_flag = 0
                for j in range(self.nn_state.num_visible):
                    if (bases_batch[i][j] != "Z"):

After Change

        for net in self.nn_state.networks:
            for par in grad_data[net].keys():
                grad[net][par] = grad_data[net][par]/float(samples_batch.shape[0])// - grad_model[net][par]/float(self.nn_state.visible_state.shape[0])
        for par in grad_data["rbm_am"].keys():
            grad["rbm_am"][par] -= grad_model["rbm_am"][par]/float(self.nn_state.visible_state.shape[0])
        return grad
        

Instances