Bottleneck

[charlotte-pel]

Hi,

I think linear should be conv. It might not change something for univariate time series, but I think it will for multivariate time series:

BreizhCrops/breizhcrops/models/InceptionTime.py

Line 50 in 4060d5c

self.bottleneck = nn.Linear(num_filters, out_features=1, bias=use_bias)

Cheers,
Charlotte.