fix the buf of the reduced SA

fairseq/modules/pyramid_layer.py

@@ -44,7 +44,7 @@ class PyramidTransformerEncoderLayer(nn.Module):
         self.quant_noise = getattr(args, 'quant_noise_pq', 0)
         self.quant_noise_block_size = getattr(args, 'quant_noise_pq_block_size', 8) or 8
         self.attn_type = getattr(args, "encoder_attention_type", "selfattn")
-        self.self_attn = self.build_self_attention(self.embed_dim, num_head, args)
+        self.self_attn = self.build_self_attention(args, self.embed_dim, num_head, att_sample_ratio)
         self.self_attn_layer_norm = LayerNorm(self.embed_dim)
         self.dropout_module = FairseqDropout(
             args.dropout, module_name=self.__class__.__name__