Open dwwx opened 4 years ago
@classmethod def loss(self, pred, label): return (pred[0] - label) * 2 @classmethod def bottom_diff(self, pred, label): diff = np.zeros_like(pred) diff[0] = 2 (pred[0] - label) return diff why use pred[0] to compute losee and buttom_diff
@classmethod def loss(self, pred, label): return (pred[0] - label) * 2 @classmethod def bottom_diff(self, pred, label): diff = np.zeros_like(pred) diff[0] = 2 (pred[0] - label) return diff why use pred[0] to compute losee and buttom_diff