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

translation model updates

Browse Source
This commit is contained in:
Mike J Innes 2017-06-28 09:15:44 -07:00
parent ea5d43ed77
commit 2be51b8c84
1 changed files with 10 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
examples/char-rnn.jl
View File

@ -1,23 +1,22 @@
using Flux using Flux
using Flux: onehot, logloss, unsqueeze
using Flux.Batches: Batch, tobatch, seqs, chunk
import StatsBase: wsample import StatsBase: wsample
nunroll = 50 nunroll = 50
nbatch = 50 nbatch = 50
getseqs(chars, alphabet) = encode(input) = seqs((onehot(ch, alphabet) for ch in input), nunroll)
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"); cd(@__DIR__)
input = readstring("shakespeare_input.txt");
alphabet = unique(input) alphabet = unique(input)
N = length(alphabet) N = length(alphabet)
train = zip(getbatches(input, alphabet), getbatches(input[2:end], alphabet)) Xs = (Batch(ss) for ss in zip(encode.(chunk(input, 50))...))
eval = tobatch.(first(drop(train, 5))) Ys = (Batch(ss) for ss in zip(encode.(chunk(input[2:end], 50))...))
model = Chain( model = Chain(
Input(N),
LSTM(N, 256), LSTM(N, 256),
LSTM(256, 256), LSTM(256, 256),
Affine(256, N), Affine(256, N),
@ -25,9 +24,10 @@ model = Chain(
m = mxnet(unroll(model, nunroll)) m = mxnet(unroll(model, nunroll))
eval = tobatch.(first.(drop.((Xs, Ys), 5)))
evalcb = () -> @show logloss(m(eval[1]), eval[2]) evalcb = () -> @show logloss(m(eval[1]), eval[2])
@time Flux.train!(m, train, η = 0.1, loss = logloss, cb = [evalcb]) # @time Flux.train!(m, zip(Xs, Ys), η = 0.001, loss = logloss, cb = [evalcb], epoch = 10)
function sample(model, n, temp = 1) function sample(model, n, temp = 1)
s = [rand(alphabet)] s = [rand(alphabet)]
@ -38,4 +38,4 @@ function sample(model, n, temp = 1)
return string(s...) return string(s...)
end end
s = sample(model[1:end-1], 100) # s = sample(model[1:end-1], 100)