Closed Sobsz closed 7 years ago
Could you please try what happens when you pull the latest changes and reinstall dependencies? I have removed the dependency on my keras branch.
Okay, I redownloaded the master branch, readded my .wav files, reinstalled the dependencies and... not much has changed, really. Here's the log:
C:\Windows\system32>pip install -r "C:\Users\Nowy Minecraft\Documents\WaveNet\re
quirements.txt"
Collecting theano~=0.9b1 (from -r C:\Users\Nowy Minecraft\Documents\WaveNet\requ
irements.txt (line 1))
Downloading Theano-0.9.0b1.zip (3.4MB)
100% |################################| 3.4MB 48kB/s
Requirement already satisfied: keras~=1.2.1 in c:\python34\lib\site-packages (fr
om -r C:\Users\Nowy Minecraft\Documents\WaveNet\requirements.txt (line 2))
Requirement already satisfied: picklable_itertools~=0.1.1 in c:\python34\lib\sit
e-packages (from -r C:\Users\Nowy Minecraft\Documents\WaveNet\requirements.txt (
line 3))
Requirement already satisfied: sacred~=0.6.10 in c:\python34\lib\site-packages (
from -r C:\Users\Nowy Minecraft\Documents\WaveNet\requirements.txt (line 4))
Requirement already satisfied: h5py~=2.6.0 in c:\python34\lib\site-packages (fro
m -r C:\Users\Nowy Minecraft\Documents\WaveNet\requirements.txt (line 5))
Collecting tqdm~=4.8.4 (from -r C:\Users\Nowy Minecraft\Documents\WaveNet\requir
ements.txt (line 6))
Downloading tqdm-4.8.4-py2.py3-none-any.whl
Requirement already satisfied: q~=2.6 in c:\python34\lib\site-packages (from -r
C:\Users\Nowy Minecraft\Documents\WaveNet\requirements.txt (line 7))
Requirement already satisfied: numpy>=1.9.1 in c:\python34\lib\site-packages (fr
om theano~=0.9b1->-r C:\Users\Nowy Minecraft\Documents\WaveNet\requirements.txt
(line 1))
Requirement already satisfied: scipy>=0.14 in c:\python34\lib\site-packages (fro
m theano~=0.9b1->-r C:\Users\Nowy Minecraft\Documents\WaveNet\requirements.txt (
line 1))
Requirement already satisfied: six>=1.9.0 in c:\python34\lib\site-packages (from
theano~=0.9b1->-r C:\Users\Nowy Minecraft\Documents\WaveNet\requirements.txt (l
ine 1))
Requirement already satisfied: pyyaml in c:\python34\lib\site-packages (from ker
as~=1.2.1->-r C:\Users\Nowy Minecraft\Documents\WaveNet\requirements.txt (line 2
))
Requirement already satisfied: wrapt in c:\python34\lib\site-packages (from sacr
ed~=0.6.10->-r C:\Users\Nowy Minecraft\Documents\WaveNet\requirements.txt (line
4))
Requirement already satisfied: docopt in c:\python34\lib\site-packages (from sac
red~=0.6.10->-r C:\Users\Nowy Minecraft\Documents\WaveNet\requirements.txt (line
4))
Installing collected packages: theano, tqdm
Found existing installation: Theano 0.8.2
Uninstalling Theano-0.8.2:
Successfully uninstalled Theano-0.8.2
Running setup.py install for theano ... done
Found existing installation: tqdm 4.11.1
Uninstalling tqdm-4.11.1:
Successfully uninstalled tqdm-4.11.1
Successfully installed theano-0.9.0b1 tqdm-4.8.4
C:\Windows\system32>cd "C:\Users\Nowy Minecraft\Documents\WaveNet\
C:\Users\Nowy Minecraft\Documents\WaveNet>python wavenet.py
Using Theano backend.
WARNING (theano.configdefaults): g++ not available, if using conda: `conda insta
ll m2w64-toolchain`
WARNING (theano.configdefaults): g++ not detected ! Theano will be unable to exe
cute optimized C-implementations (for both CPU and GPU) and will default to Pyth
on implementations. Performance will be severely degraded. To remove this warnin
g, set Theano flags cxx to an empty string.
INFO - wavenet - Running command 'main'
WARNING - wavenet - No observers have been added to this run
INFO - wavenet - Started
Configuration (modified, added, typechanged):
batch_size = 16
data_dir = 'data'
data_dir_structure = 'flat'
debug = False
desired_sample_rate = 4410
dilation_depth = 9
early_stopping_patience = 20
final_l2 = 0
fragment_length = 1152
fragment_stride = 128
keras_verbose = 1
learn_all_outputs = True
nb_epoch = 1000
nb_filters = 256
nb_output_bins = 256
nb_stacks = 1
predict_initial_input = None
predict_seconds = 1
predict_use_softmax_as_input = False
random_train_batches = False
randomize_batch_order = True
res_l2 = 0
run_dir = None
sample_argmax = False
sample_temperature = 1.0
seed = 393618564
test_factor = 0.1
train_only_in_receptive_field = True
train_with_soft_target_stdev = None
use_bias = False
use_skip_connections = True
use_ulaw = True
optimizer:
decay = 0.0
epsilon = None
lr = 0.001
momentum = 0.9
nesterov = True
optimizer = 'sgd'
INFO - main - Running with seed 393618564
INFO - main - Loading data...
0%| | 0/1 [00:00<?, ?it/s]
I would suggest to try a UNIX based OS to work with this, I think tensorflow has a nice tutorial to set up docker on Windows. Take a look there :).
On Thu, 16 Feb 2017, 19:13 Simon, notifications@github.com wrote:
Okay, I redownloaded the master branch, readded my .wav files, reinstalled the dependencies and... not much has changed, really. Here's the log:
C:\Windows\system32>pip install -r "C:\Users\Nowy Minecraft\Documents\WaveNet\re quirements.txt" Collecting theano~=0.9b1 (from -r C:\Users\Nowy Minecraft\Documents\WaveNet\requ irements.txt (line 1)) Downloading Theano-0.9.0b1.zip (3.4MB) 100% |################################| 3.4MB 48kB/s Requirement already satisfied: keras~=1.2.1 in c:\python34\lib\site-packages (fr om -r C:\Users\Nowy Minecraft\Documents\WaveNet\requirements.txt (line 2)) Requirement already satisfied: picklable_itertools~=0.1.1 in c:\python34\lib\sit e-packages (from -r C:\Users\Nowy Minecraft\Documents\WaveNet\requirements.txt ( line 3)) Requirement already satisfied: sacred~=0.6.10 in c:\python34\lib\site-packages ( from -r C:\Users\Nowy Minecraft\Documents\WaveNet\requirements.txt (line 4)) Requirement already satisfied: h5py~=2.6.0 in c:\python34\lib\site-packages (fro m -r C:\Users\Nowy Minecraft\Documents\WaveNet\requirements.txt (line 5)) Collecting tqdm~=4.8.4 (from -r C:\Users\Nowy Minecraft\Documents\WaveNet\requir ements.txt (line 6)) Downloading tqdm-4.8.4-py2.py3-none-any.whl Requirement already satisfied: q~=2.6 in c:\python34\lib\site-packages (from -r C:\Users\Nowy Minecraft\Documents\WaveNet\requirements.txt (line 7)) Requirement already satisfied: numpy>=1.9.1 in c:\python34\lib\site-packages (fr om theano~=0.9b1->-r C:\Users\Nowy Minecraft\Documents\WaveNet\requirements.txt (line 1)) Requirement already satisfied: scipy>=0.14 in c:\python34\lib\site-packages (fro m theano~=0.9b1->-r C:\Users\Nowy Minecraft\Documents\WaveNet\requirements.txt ( line 1)) Requirement already satisfied: six>=1.9.0 in c:\python34\lib\site-packages (from theano~=0.9b1->-r C:\Users\Nowy Minecraft\Documents\WaveNet\requirements.txt (l ine 1)) Requirement already satisfied: pyyaml in c:\python34\lib\site-packages (from ker as~=1.2.1->-r C:\Users\Nowy Minecraft\Documents\WaveNet\requirements.txt (line 2 )) Requirement already satisfied: wrapt in c:\python34\lib\site-packages (from sacr ed~=0.6.10->-r C:\Users\Nowy Minecraft\Documents\WaveNet\requirements.txt (line 4)) Requirement already satisfied: docopt in c:\python34\lib\site-packages (from sac red~=0.6.10->-r C:\Users\Nowy Minecraft\Documents\WaveNet\requirements.txt (line 4)) Installing collected packages: theano, tqdm Found existing installation: Theano 0.8.2 Uninstalling Theano-0.8.2: Successfully uninstalled Theano-0.8.2 Running setup.py install for theano ... done Found existing installation: tqdm 4.11.1 Uninstalling tqdm-4.11.1: Successfully uninstalled tqdm-4.11.1 Successfully installed theano-0.9.0b1 tqdm-4.8.4
C:\Windows\system32>cd "C:\Users\Nowy Minecraft\Documents\WaveNet\
C:\Users\Nowy Minecraft\Documents\WaveNet>python wavenet.py Using Theano backend. WARNING (theano.configdefaults): g++ not available, if using conda:
conda insta ll m2w64-toolchain
WARNING (theano.configdefaults): g++ not detected ! Theano will be unable to exe cute optimized C-implementations (for both CPU and GPU) and will default to Pyth on implementations. Performance will be severely degraded. To remove this warnin g, set Theano flags cxx to an empty string. INFO - wavenet - Running command 'main' WARNING - wavenet - No observers have been added to this run INFO - wavenet - Started Configuration (modified, added, typechanged): batch_size = 16 data_dir = 'data' data_dir_structure = 'flat' debug = False desired_sample_rate = 4410 dilation_depth = 9 early_stopping_patience = 20 final_l2 = 0 fragment_length = 1152 fragment_stride = 128 keras_verbose = 1 learn_all_outputs = True nb_epoch = 1000 nb_filters = 256 nb_output_bins = 256 nb_stacks = 1 predict_initial_input = None predict_seconds = 1 predict_use_softmax_as_input = False random_train_batches = False randomize_batch_order = True res_l2 = 0 run_dir = None sample_argmax = False sample_temperature = 1.0 seed = 393618564 test_factor = 0.1 train_only_in_receptive_field = True train_with_soft_target_stdev = None use_bias = False use_skip_connections = True use_ulaw = True optimizer: decay = 0.0 epsilon = None lr = 0.001 momentum = 0.9 nesterov = True optimizer = 'sgd' INFO - main - Running with seed 393618564 INFO - main - Loading data... 0%| | 0/1 [00:00<?, ?it/s]— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub https://github.com/basveeling/wavenet/issues/25#issuecomment-280411923, or mute the thread https://github.com/notifications/unsubscribe-auth/AAgxjxzuWT1lTlTSIaU7uAS7y6PuR9lrks5rdJGugaJpZM4Lts8a .
After a few days of troubleshooting, I got everything to work (read: not throw exceptions). However, while the script runs fine, it seems to get stuck on the "loading data" step. I've let it run for over an hour and it doesn't seem to be making any progress. Task Manager shows it's using quite a bit of CPU and RAM, however the latter doesn't seem to change during runtime. I'm using Win7 Home Premium 64-bit with Python 3.4.3 32-bit. NumPy is 1.11.3+mkl, SciPy is 0.18.1 (upgraded from 0.16.1 in .exe), Theano is 0.8.2, Keras is 1.2.1 (basveeling's branch lacks at least one crucial method). Log of stdout: