Meaning of batch sizes for RNNs/LSTMs and reasons for padding

I’ve got a two conceptual questions about RNNs, particularly LSTMs, which I just can’t figure out on my own or with the tutorials I find on the internet. I would really appreciate if you could help me with the following:

  1. If I understand correctly, the states learned within a LSTM are only relevant for one sequence. So, for the next sequence the states are being “relearned” due to $s_{t}=f(Ux_{t} + Ws_{t-1})$ with x being the input at timestep t, s being the state at timestep t and U and W being the matrices that are learned. Is there any good reason why you should use larger batch sizes than 1 with RNNs/LSTMs especially? I know the differences between stoachastic gradient descent, batch gradient descent and Mini-batch gradient descent, but not why the latter two should be preferred over the first one in RNNs/LSTMs.
  2. Why do you need the same sequence lengths within a batch, i.e. why is padding needed? The states are calculated for each sequence separately, so I don’t see a reason for this. Does the backprop through time need the same number of states for each sequence, when it’s being executed after a batch?

Technical reasons against chocolate-coated shortbread

It is possible to find chocolate-coated shortbread (as in, the Scottish biscuit made of flour, sugar and butter, coated with chocolate) as a product in the UK, but it’s very rare. Chocolate-coated shortcake — which is similar, but different — is much more common. “Millionaire’s shortbread” is easy to find, but it’s not completely coated in chocolate and has a thick caramel layer, making it quite a different product.

Given that shortbread is very popular — and, presumably, very cheap to make — and that chocolate is very popular, I’m led to believe that there must be some technical reason why it’s so hard to find. Does anyone have any ideas?

My only thought is that it may be due to the fat interface between the layers (i.e., butter against cocoa butter) not playing together nicely, reducing shelf-stability.

Reasons for being vegetarian or vegan other than ethical reasons?

I met a lot of people that, even though they have strong reasons for being vegan or vegetarian, or even work saving or helping animals, etc. have failed to follow their principles and ended up including meat again.

Almost every vegetarian or vegan person I get to know has moved to vegetarianism for the same reasons I did (I’m moving in as well):

  • Ethical reasons (i.e., not eating other animals)
  • Trying to stop animal suffering (defending animals rights).

But, are there reasons besides the ones listed above?

I’m not saying they’re not great reasons, which I think they are, just trying to find out all other reasons.

Are large halo orbits around L₁’s and L₂’s preferred over small orbits for reasons other than geometry?

There have been many examples of the placement of satellites in orbits around Lagrange points, most have been sun-earth and earth-moon $L_1$ and $L_2$ due to their proximity to earth. In each case both bodies and both points fall on a straight line, so to have the right combination of non-blocked line-of-sight to earth for communications, safe angular separation from the sun, and non-blocked line-of-sight to the sun for solar power, the size and shape orbits about the Lagrange points are carefully optimized.

Besides those reasons, are there orbital mechanical or other reasons why placement in orbit around L₁ and L₂ would be preferred over placement near the points themselves? Does it require significantly more delta-v to get there (shortening station keeping lifetime)? Would station keeping be more difficult or costly (delta-v) near the point than it would be in a larger orbit about the point (excluding communications issues)?

Help with reasons for rebellions against my oligarchic/Philosopher kings inspired empire [closed]

The government of this (fantasy) world is made up of a small group, randomly chosen at birth and representing maybe 0.5% of the population. The group are tutored in their homes with their families for 5 years, then retain some contact whilst in specialist academies until the age of 18 when they assume full voting rights as part of the Assembly, and travel to various government buildings/courts to vote in the affairs of the state.
I am aware of the problems any empire may have (religious differences, clashing cultural identities etc.), but I want opinions on what other reasons groups might rebel against this empire. For example, the one reason I already have is that it is slow (takes 18 years) to add representatives of other cultures. It also occurs to me that gerontocratic tribes may have trouble excepting 18 year olds as their representatives, as to start with the empire likes to work with representatives of the old structure.

Which reasons can break a mountpoint in mp3fs?

I use mp3fs on a samba share (openmediavault). When I copy from that mp3fs mountpoint on the share (via terminal or via file manager) usually the process fails at some point thereby crashing the mount point. I have to SSH into the share and unmount and re-mount the mp3fs mountpoint. My fstab looks like this:

mp3fs#/sharedfolders/RAID6/Musik /sharedfolders/RAID6/mp3ben192 fuse allow_other,ro,bitrate=192 0 0

Copying the files from the mp3fs to another directory on the share itself (via terminal and SSH) works fine. So it has to be a problem of the combination of mp3fs and cifs/samba.

The error I get when copying via samba is permission related. I have tried a lot of options of cifs.mount to make sure it is not related to the way I mount the samba. I now make a forceuser in the smb.conf on the share so I am pretty sure that the right user is used to read the files, but even with guest mode it would fail.

It is important to notice that it does not fail with all files, but just certain ones. All files have the same permissions, user and group of course.