Update modeling_super_linear.py

modeling_super_linear.py

@@ -347,6 +347,7 @@ class SparseNoisyMoE(nn.Module):
             x_0 = x
             
         self.gate_outputs = self.gating_network(x_0)
+        print(self.gate_outputs.shape)
 
         if not self.training:
             self.gate_outputs = self.gate_outputs / self.moe_temp