bquast / rnn

Recurrent Neural Networks in R
https://qua.st/rnn
73 stars 28 forks source link

correct error #16

Closed MochizukiShinichi closed 7 years ago

MochizukiShinichi commented 7 years ago

Binary sequence should be small to large from left to right.

coveralls commented 7 years ago

Coverage Status

Coverage remained the same at 33.917% when pulling 43a14538f9d48b8515ebc2c1f227187dd05c8592 on MochizukiShinichi:master into 51341bc757fa327963b43e06cbe8d4e67647a3b1 on bquast:master.

bquast commented 7 years ago

@MochizukiShinichi indeed R prints sequences out of memory from left to right, which is opposite to the way we would write it. In the original code, I did rev() the order (see commit history), but as it turns out, this means you the RNN also so feed it the wrong way around, which needlessly complicates the system. It is, therefore simpler to print it the wrong way around.

I will be removing these functions in a future release since they were more relevant to the initial PoW then to rnn in its current state and then simply define them in the examples and vignettes.