siemanko

"""Short and sweet LSTM implementation in Tensorflow.

Motivation:
When Tensorflow was released, adding RNNs was a bit of a hack - it required
building separate graphs for every number of timesteps and was a bit obscure
to use. Since then TF devs added things like `dynamic_rnn`, `scan` and `map_fn`.
Currently the APIs are decent, but all the tutorials that I am aware of are not
making the best use of the new APIs.

Advantages of this implementation:

mrange

[F#] Implementing Coroutines (async/await) using continuations

Coroutines or async/await is not a new concept as it was named in 1958 by Melvin Conway. Coroutines had support in languages such as Simula(1962), Smalltalk(1972) and Modula-2(1978) but I think one can argue that coroutines came into mainstream with C# 5(2012) when Microsoft added async/await to C#.

The problem with subroutines

A thread can only execute a single subroutine at a time as a subroutine starts and then runs to completion. In a server environment subroutines can be problematic as in order to service connections concurrently we would usually would start a thread per connection, this was the idiom a few years ago. A thread depending on the operating system and settings needs about 1 MiB of user stack space meaning 1,024 threads needs 1 GiB of for just user stack space. In addition there's the cost of kernel stack space, book-keeping and scheduling of threads. This drives cost of VM/hardwa