Make the forward function, which performs processing of x, more concise

[kounkou]

Description

I am currently reading the code and noticed we could improve the code to be more concise.
My proposal is to go from :

def forward(self, x):
        x = x.repeat(1, self._options["num_cnns"], 1)
        if self._options["num_fc"] <= 2:
            outs = self.linears(x)
        else:
            outs = self.linears(x)
            for i in range(len(self.linears_bg)):
                outs = self.linears_bg[i](outs)
            outs = self.last_linear(outs)
        out = self.final(outs)

to :

def forward(self, x):
    x = x.repeat(1, self._options["num_cnns"], 1)
    outs = self.linears(x)

    if self._options["num_fc"] > 2:
        for layer in self.linears_bg:
            outs = layer(outs)
        outs = self.last_linear(outs)
    
    return self.final(outs)

Testing

Currently, I just made sure the train can still proceed, since the update affects the networks only.
Well let me know if pull requests are welcome...