Skip to content

Instantly share code, notes, and snippets.

View ozancaglayan's full-sized avatar

Ozan Caglayan ozancaglayan

View GitHub Profile
losses = [57.4, 50.7, 40.9, 39.6 ,37.9, 37.5, 36.3, 35.7, 36.5, 35.13, 36.33, 34.37, 34.78, 34.67, 34.44, 35.2, 35.66, 32.47, 34.6, 34.7, 35.14, 34.5]
import numpy as np
vhist = []
patience = 10
bad_c = 0
use_bleu = False
for i in range(len(losses)):
loss = losses[i]
#!/usr/bin/env python
import sys
import time
def model():
print "model"
try:
while 1:
print "loop"
time.sleep(2)
#!/usr/bin/env python
import sys
import numpy as np
import time
import theano
import theano.tensor as T
@ozancaglayan
ozancaglayan / index.py
Created December 10, 2015 17:51
advanced indexing
# pp: (batch, sequence_step, target vocabulary probabilities)
# yy: (batch, sequence_step's true label)
# Soru: pp[yy] gibi dogru yerlerden dogru olasiliklari nasi cekebilirim?
In [211]: pp.shape
Out[211]: (256, 33, 20004)
In [212]: yy.shape
Out[212]: (256, 33)
#!/usr/bin/env python
from keras.layers.core import Dense
from keras.models import Sequential
from keras.optimizers import RMSprop, Adadelta, Adagrad, Adam, SGD
from keras.callbacks import Callback
import numpy as np
class Test(Callback):
@ozancaglayan
ozancaglayan / testarch3.log
Created November 13, 2015 15:21
testarch3 log
Creating a new machine
- initializing projections with random values in the range 0.1
- initializing weights with random values in the range 0.1
Initializing Nvidia GPU card
- found 8 cards:
0: Tesla K40c with 15 CPUs x 192 threads running at 0.74 Ghz, 11519 MBytes of memory, use -arch=sm_35, utilization 0%
1: Tesla K40c with 15 CPUs x 192 threads running at 0.74 Ghz, 11519 MBytes of memory, use -arch=sm_35, utilization 0%
2: Tesla K40c with 15 CPUs x 192 threads running at 0.74 Ghz, 11519 MBytes of memory, use -arch=sm_35, utilization 0%
3: Tesla K40c with 15 CPUs x 192 threads running at 0.74 Ghz, 11519 MBytes of memory, use -arch=sm_35, utilization 0%
4: Tesla K40c with 15 CPUs x 192 threads running at 0.74 Ghz, 11519 MBytes of memory, use -arch=sm_35, utilization 0%
@ozancaglayan
ozancaglayan / repeatbug.patch
Created November 13, 2015 14:29
repeat-bug-logs
--- c384-ce12288-pe1536-h256-bs256-lr0.03.log 2015-10-29 22:28:40.539747323 +0100
+++ c384-ce12288-pe1536-h256-bs256-lr0.03-repeatbug.log 2015-11-13 15:41:20.820063590 +0100
@@ -12,6 +12,7 @@
6: Tesla K40c with 15 CPUs x 192 threads running at 0.74 Ghz, 11519 MBytes of memory, use -arch=sm_35, utilization 0%
7: Tesla K40c with 15 CPUs x 192 threads running at 0.74 Ghz, 11519 MBytes of memory, use -arch=sm_35, utilization 0%
- using 8 devices in parallel: 0 1 2 3 4 5 6 7
+ - repeating these 1 machine(s) 32 times
#### GPU allocate local data for 8 GPU
#### GPU 0: use local data_in from MachSplit
#### GPU 1: allocate local data_in=0x2162a0000

Keybase proof

I hereby claim:

  • I am ozancaglayan on github.
  • I am ozancaglayan (https://keybase.io/ozancaglayan) on keybase.
  • I have a public key whose fingerprint is 2C3B 4DA8 7E9B B14C 2B6B CB51 E96D AF7F F229 6D8C

To claim this, I am signing this object: