Initial commit: FunASR Speech Recognition Toolkit

Add complete FunASR codebase including models, runtime, and documentation.

funasr/models/llm_asr_nar/adaptor.py

@@ -0,0 +1,44 @@
+import torch
+import torch.nn as nn
+
+from funasr.register import tables
+
+
+@tables.register("adaptor_classes", "Linear")
+class Linear(nn.Module):
+    def __init__(self, downsample_rate, encoder_dim, llm_dim, ffn_dim: int = 2048, **kwargs):
+        """Initialize Linear.
+        
+            Args:
+                downsample_rate: TODO.
+                encoder_dim: Size/dimension parameter.
+                llm_dim: Size/dimension parameter.
+                ffn_dim: Size/dimension parameter.
+                **kwargs: Additional keyword arguments.
+            """
+        super().__init__()
+        self.k = downsample_rate
+        self.encoder_dim = encoder_dim
+        self.llm_dim = llm_dim
+        self.linear1 = nn.Linear(self.encoder_dim * self.k, ffn_dim)
+        self.relu = nn.ReLU()
+        self.linear2 = nn.Linear(ffn_dim, self.llm_dim)
+
+    def forward(self, x):
+        """Forward pass for training.
+        
+            Args:
+                x: TODO.
+            """
+        batch_size, seq_len, dim = x.size()
+        num_frames_to_discard = seq_len % self.k
+        if num_frames_to_discard > 0:
+            x = x[:, :-num_frames_to_discard, :]
+        seq_len = x.size(1)
+
+        x = x.contiguous()
+        x = x.view(batch_size, seq_len // self.k, dim * self.k)
+        x = self.linear1(x)
+        x = self.relu(x)
+        x = self.linear2(x)
+        return x