TastyPancakes/Flux.jl
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

magic constants

Browse Source
This commit is contained in:
Mike J Innes 2016-12-15 21:07:07 +00:00
parent 752c4e39e8
commit d3e0f455c2
1 changed files with 6 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
examples/char-rnn.jl
View File

@ -1,8 +1,11 @@
using Flux using Flux
import StatsBase: wsample import StatsBase: wsample
getseqs(chars, alphabet) = sequences((onehot(Float32, char, alphabet) for char in chars), 50) nunroll = 50
getbatches(chars, alphabet) = batches((getseqs(part, alphabet) for part in chunk(chars, 50))...) nbatch = 50
getseqs(chars, alphabet) = sequences((onehot(Float32, char, alphabet) for char in chars), nunroll)
getbatches(chars, alphabet) = batches((getseqs(part, alphabet) for part in chunk(chars, nbatch))...)
input = readstring("$(homedir())/Downloads/shakespeare_input.txt") input = readstring("$(homedir())/Downloads/shakespeare_input.txt")
alphabet = unique(input) alphabet = unique(input)
@ -17,7 +20,7 @@ model = Chain(
Affine(256, N), Affine(256, N),
softmax) softmax)
m = tf(unroll(model, 50)); m = tf(unroll(model, nunroll))
@time Flux.train!(m, Xs, Ys, η = 0.1, epoch = 1) @time Flux.train!(m, Xs, Ys, η = 0.1, epoch = 1)