Collate function for TSData objects, returning a dictionary with only present fields.

dump the contents of string to a file called filename

nn.AdaptiveAvgPool layer for `ndim`

nn.AvgPool layer for `ndim`

nn.MaxPool layer for `ndim`

input shape: (bs,channels,seq) + optional static (bs,feat) selected encoders e.g. NoEncoder also accept (bs,channels,freq,seq) for th

input shape: (bs,channels,seq) + optional static (bs,feat) selected encoders e.g. NoEncoder also accept (bs,channels,freq,seq) for th

if target_dim is not None, the output dimension should be target_dim input shape: bs, channels output shape: bs, feat

input shape: bs, seq, feat output shape: bs, seq, feat

RNN Encoder is actually just a conv encoder

L: Maximum length; this module computes an SSM kernel of length L w: (N) p: (r, N) low-rank correction to A q: (r, N)

tie: tie dropout mask across sequence lengths (Dropout1d/2d/3d)

Output will be 2*sz or 2 if sz is None