Support unbalanced structures

egs/libriheavy/LM/zipformer1/subformer.py

@@ -289,6 +289,25 @@ class Subformer(EncoderInterface):
         # d = self.output_downsampling_factor
         # lengths = (x_lens + d - 1) // d
 
+
+        # The next code block will only run in the case of "unbalanced" structures, e.g.
+        # if structure == "S(S(S)S", where there are unmatched right-parentheses.
+        cur_indexes = None
+        while len(downsample_info) > 0:
+            indexes, weights, x_orig = downsample_info.pop()
+            if cur_indexes is not None:
+                # keep only a subset of the indexes and weights, corresponding
+                # to later downsampling operations.
+                indexes = torch.gather(indexes, dim=1, index=cur_indexes)
+                weights = torch.gather(weights, dim=1, index=cur_indexes)
+
+            cur_indexes = indexes
+
+            x_lens = (weights != 0).sum(dim=1)
+            x_orig = convert_num_channels(x_orig, x.shape[-1])
+            x_orig, x = LearnedDownsamplingModule.apply_weights(x_orig, x, indexes, weights)
+
+
         return x, x_lens
 
     def _get_attn_offset(self, x: Tensor, src_key_padding_mask: Optional[Tensor]) -> Optional[Tensor]:
@@ -1063,7 +1082,7 @@ class LearnedDownsamplingModule(nn.Module):
             x_orig: (seq_len, batch_size, num_channels)
             x: (seq_len_reduced, batch_size, num_channels)
           indexes: (batch_size, seq_len_reduced), contains original frame indexes
-          weights: optional tensor
+          weights: optional tensor of shape (batch_size, seq_len_reduced)
 
         Downsamples x via indexing with the indexes obtained from the
         forward() function.
@@ -1098,97 +1117,34 @@ class LearnedDownsamplingModule(nn.Module):
         # add in x_orig in the frames that were not originally kept.
         return ans + x_orig * orig_x_weight.t().unsqueeze(-1)
 
-
+    @staticmethod
-class DownsampledSubformerEncoder(nn.Module):
+    def apply_weights(x_orig: Tensor, x: Tensor, indexes: Tensor,
-    """
+                      weights: Optional[Tensor] = None) -> Tuple[Tensor, Tensor]:
-    DownsampledSubformerEncoder is a zipformer encoder stack possibly evaluated at a reduced
+        """
-    frame rate, after convolutional downsampling, and then upsampled again at the output, and combined
+        Downsamples x_orig to have the same shape as x and applies the weights,
-    with the origin input, so that the output has the same shape as the input.
+        returning interpolated x and downsampled x_orig.  This is similar to
-    """
+        `upsample`, but is for the case where you don't want to keep the frames
-    def __init__(self,
+        that were not sampled.
-                 encoders: List[nn.Module],
-                 input_num_channels: int,
-                 downsample: int):
-        super(DownsampledSubformerEncoder, self).__init__()
-        if downsample != 1:
-            self.downsampler = LearnedDownsamplingModule(input_num_channels,
-                                                         downsample)
-
-        self.encoders = nn.ModuleList(encoders)
-
-        self.out_combiner = BypassModule(self.embed_dim(),
-                                         straight_through_rate=0.0)
-
-    def embed_dim(self):  # return output embed_dim which is max dim.
-        return max(e.embed_dim() for e in self.encoders)
-
-    def forward(self,
-                src: Tensor,
-                pos_emb: Tensor,
-                attn_offset: Tensor,
-                feature_mask: Union[Tensor, float] = 1.0,
-                memory: Optional[Tensor] = None,
-                memory_key_padding_mask: Optional[Tensor] = None,
-    ) -> Tuple[Tensor, Tensor]:
-        r"""Downsample, go through encoder, upsample.
 
         Args:
-            src: the sequence to the encoder (required): shape (seq_len, batch_size, embedding_dim).
+            x_orig: (seq_len, batch_size, num_channels)
-            pos_emb: the positional embedding, of shape (batch_size, seq_len, seq_len, pos_dim)
+            x: (seq_len_reduced, batch_size, num_channels)
-            feature_mask: something that broadcasts with src, that we'll multiply `src`
+          indexes: (batch_size, seq_len_reduced), contains original frame indexes
-               by at every layer: if a Tensor, likely of shape (seq_len, batch_size, embedding_dim)
+          weights: optional tensor of shape (batch_size, seq_len_reduced)
-            attn_offset: the attention offset, added to scores for attention of shape
-                 (batch_size, seq_len, seq_len) or (seq_len, seq_len),
-                 interpreted as (batch_size, tgt_seq_len, src_seq_len) or (tgt_seq_len, src_seq_len).
-            memory:  optionally, the memory embeddings of shape (memory_len, batch_size, memory_dim)
-            memory_key_padding_mask: optionally the mask for padding of memory input (for source-
-                attention), of shape  (batch_size, memory_len); True means
-                 masked position.  May be None.
 
-        Returns: a Tensor with the same shape as src.
+        Returns (x_orig, x) after the downsampling and interpolation, of shapes
+           both (seq_len_reduced, batch_size, num_channels).
         """
-        src_orig = src
+        (seq_len, batch_size, num_channels) = x_orig.shape
-        if hasattr(self, 'downsampler'):
+        weights = 1.0 if weights is None else weights.t().unsqueeze(-1)
-            indexes, weights, src = self.downsampler(src)
 
-            pos_emb = self.downsampler.downsample_pos_emb(pos_emb, indexes)
+        indexes = indexes.t().unsqueeze(-1).expand(-1, batch_size, num_channels)
+        # indexes now: (seq_len_reduced, batch_size, num_channels)
+        x_orig = torch.gather(x_orig, dim=0, index=indexes)
 
-            attn_offset = self.downsampler.downsample_attn_offset(attn_offset,
+        x = x * weights  +  x_orig * (1.0 - weights)
-                                                                  indexes,
-                                                                  weights)
-        outputs = [ src ]
 
-        for encoder in self.encoders:
+        return x_orig, x
-            src = encoder(
-                src,
-                pos_emb,
-                attn_offset=attn_offset,
-                memory=memory,
-                memory_key_padding_mask=memory_key_padding_mask,
-            )
-            outputs.append(src)
-
-        def get_full_dim_output():
-            num_encoders = len(outputs)
-            output_dim = max(o.shape[-1] for o in outputs)
-            output_pieces = [ outputs[-1] ]
-            cur_dim = outputs[-1].shape[-1]
-            for i in range(num_encoders - 2, -1, -1):
-                d = outputs[i].shape[-1]
-                if d > cur_dim:
-                    this_output = outputs[i]
-                    output_pieces.append(this_output[..., cur_dim:d])
-                    cur_dim = d
-            assert cur_dim == output_dim
-            return torch.cat(output_pieces, dim=-1)
-
-        src = get_full_dim_output()
-        src_orig = convert_num_channels(src_orig, src.shape[-1])
-
-        if hasattr(self, 'downsampler'):
-            src = self.downsampler.upsample(src_orig, src, indexes, weights)
-
-        return self.out_combiner(src_orig, src)
 
 
 
@@ -1881,6 +1837,7 @@ def _test_zipformer_main(causal: bool = False):
     memory_dim = 100
 
     c = Subformer(
+        structure = "S(S)S" if causal else "S(S(S",
         encoder_dim=(64, 96, 64),
         num_heads=(4, 4, 8),
         causal=causal,