The meaning of comments in the code.

[fairy-of-9]

Hi! Thanks for your share code.
I'm studying with your paper with code.
I have a question about comments in your code.

EncoderMemNN in Mem2Seq.py

    def forward(self, story):
        story = story.transpose(0,1)
        story_size = story.size() # b * m * 3 
        if self.unk_mask:
            if(self.training):
                ones = np.ones((story_size[0],story_size[1],story_size[2]))
                rand_mask = np.random.binomial([np.ones((story_size[0],story_size[1]))],1-self.dropout)[0]
                ones[:,:,0] = ones[:,:,0] * rand_mask
                a = Variable(torch.Tensor(ones))
                if USE_CUDA: a = a.cuda()
                story = story*a.long()
        u = [self.get_state(story.size(0))]
        for hop in range(self.max_hops):
            embed_A = self.C[hop](story.contiguous().view(story.size(0), -1).long()) # b * (m * s) * e
            embed_A = embed_A.view(story_size+(embed_A.size(-1),)) # b * m * s * e
            m_A = torch.sum(embed_A, 2).squeeze(2) # b * m * e

What does b, m, s, e, 3 mean?

Thank you!