Merge pull request #747 from AIStream-Peelout/crossformer_fixes

Crossformer fixes

flood_forecast/transformer_xl/cross_former.py

@@ -19,6 +19,7 @@ def __init__(
         e_layers=3,
         dropout=0.0,
         baseline=False,
+        n_targs=None,
         device=torch.device("cuda:0"),
     ):
         """Crossformer: Transformer Utilizing Cross-Dimension Dependency for Multivariate Time Series Forecasting.
@@ -57,6 +58,7 @@ def __init__(
         self.out_len = forecast_length
         self.seg_len = seg_len
         self.merge_win = win_size
+        self.n_targs = n_time_series if n_targs is None else n_targs
 
         self.baseline = baseline
 
@@ -126,7 +128,9 @@ def forward(self, x_seq: torch.Tensor):
         )
         predict_y = self.decoder(dec_in, enc_out)
 
-        return base + predict_y[:, : self.out_len, :]
+        result = base + predict_y[:, : self.out_len, :]
+        res = result[:, :, :self.n_targs]
+        return res
 
 
 class SegMerging(nn.Module):

flood_forecast/transformer_xl/multi_head_base.py

@@ -5,7 +5,7 @@
 
 
 class MultiAttnHeadSimple(torch.nn.Module):
-    """A simple multi-head attention model inspired by Vaswani et al."""
+    """A simple multi-head attention model inspired by Vas.,wani et al."""
 
     def __init__(
             self,

setup.py

@@ -16,7 +16,7 @@
 
 setup(
     name='flood_forecast',
-    version='1.001dev',
+    version='1.0001dev',
     packages=[
         'flood_forecast',
         'flood_forecast.transformer_xl',