[converter] set hybrid_q_type for int16 (#363)

tinynn/converter/base.py

@@ -214,6 +214,7 @@ def __init__(
         elif hybrid_quantize_weight_type == 'int8':
             self.hybrid_q_type = np.int8
         elif hybrid_quantize_weight_type == 'int16':
+            self.hybrid_q_type = np.int16
             if self.hybrid:
                 raise AttributeError('Hybrid kernels supports int8 and uint8 only')