add reverse weight

paddlespeech/s2t/exps/u2/model.py

@@ -520,6 +520,7 @@ class U2Tester(U2Trainer):
             infer_model.ctc_activation, input_spec=input_spec)
 
         ######################### infer_model.forward_attention_decoder ########################
+        reverse_weight = 0.3
         input_spec = [
             # hyps, (B, U)
             paddle.static.InputSpec(shape=[None, None], dtype='int64'),
@@ -527,7 +528,8 @@ class U2Tester(U2Trainer):
             paddle.static.InputSpec(shape=[None], dtype='int64'),
             # encoder_out, (B,T,D)
             paddle.static.InputSpec(
-                shape=[batch_size, None, model_size], dtype='float32')
+                shape=[batch_size, None, model_size], dtype='float32'),
+            reverse_weight
         ]
         infer_model.forward_attention_decoder = paddle.jit.to_static(
             infer_model.forward_attention_decoder, input_spec=input_spec)

paddlespeech/s2t/models/u2/u2.py

@@ -706,7 +706,7 @@ class U2BaseModel(ASRInterface, nn.Layer):
             hyps: paddle.Tensor,
             hyps_lens: paddle.Tensor,
             encoder_out: paddle.Tensor,
-            reverse_weight: float=0.0, ) -> paddle.Tensor:
+            reverse_weight: float=0.0) -> paddle.Tensor:
         """ Export interface for c++ call, forward decoder with multiple
             hypothesis from ctc prefix beam search and one encoder output
         Args: