Creates tensor in target device to avoid high CPU occupation. (#695)

funasr/modules/embedding.py

@@ -393,8 +393,9 @@ class SinusoidalPositionEncoder(torch.nn.Module):
     def encode(self, positions: torch.Tensor = None, depth: int = None, dtype: torch.dtype = torch.float32):
         batch_size = positions.size(0)
         positions = positions.type(dtype)
-        log_timescale_increment = torch.log(torch.tensor([10000], dtype=dtype)) / (depth / 2 - 1)
-        inv_timescales = torch.exp(torch.arange(depth / 2).type(dtype) * (-log_timescale_increment))
+        device = positions.device
+        log_timescale_increment = torch.log(torch.tensor([10000], dtype=dtype, device=device)) / (depth / 2 - 1)
+        inv_timescales = torch.exp(torch.arange(depth / 2, device=device).type(dtype) * (-log_timescale_increment))
         inv_timescales = torch.reshape(inv_timescales, [batch_size, -1])
         scaled_time = torch.reshape(positions, [1, -1, 1]) * torch.reshape(inv_timescales, [1, 1, -1])
         encoding = torch.cat([torch.sin(scaled_time), torch.cos(scaled_time)], dim=2)
@@ -402,7 +403,7 @@ class SinusoidalPositionEncoder(torch.nn.Module):
 
     def forward(self, x):
         batch_size, timesteps, input_dim = x.size()
-        positions = torch.arange(1, timesteps+1)[None, :]
+        positions = torch.arange(1, timesteps+1, device=x.device)[None, :]
         position_encoding = self.encode(positions, input_dim, x.dtype).to(x.device)
 
         return x + position_encoding