Closed jimmyahacker closed 6 years ago
Fix a typo in parzen_ll.py The following is the result using my instruction on how to reproduce the original GAN and wish this would help others
parzen_ll.py
Using gpu device 0: Tesla K80 (CNMeM is disabled, CuDNN not available) /scratch/sj2363/hsn/attack_generate/original_gan_reproduction/pylearn2/pylearn2/space/__init__.py:895: FutureWarning: Conversion of the second argument of issubdtype from `complex` to `np.complexfloating` is deprecated. In future, it will be treated as `np.complex128 == np.dtype(complex).type`. return np.issubdtype(dtype, np.complex) Using Sigma: 0.01 0 0.5437917709350586 -249610.04500921787 10 0.4127103198658336 -254864.56703336554 20 0.4046180702391125 -256314.53794820592 30 0.40232555327876923 -256196.53144318963 40 0.4012015796289211 -255919.3469756813 50 0.4002548199073941 -255600.8114277227 60 0.3996614237300685 -255690.43841213794 70 0.3994017889801885 -256184.89072224603 80 0.3990846445531021 -255830.50266160985 90 0.3988482244722136 -255731.5973735723 Log-Likelihood of test set = -256075.399815, se: 791.576879859 0 0.39833712577819824 -255978.92391546787 10 0.3963735753839666 -253777.21132313827 20 0.39686526571001324 -253051.69069002144 30 0.3968393264278289 -253648.935089863 40 0.39672014771438224 -253051.69475388253 50 0.396716692868401 -252101.05537533303 60 0.39688189303288696 -253178.86357351087 70 0.39701688121741924 -253049.50480345203 80 0.3969559728363414 -252923.0343543182 90 0.3969241472390982 -253003.14902707504 Log-Likelihood of valid set = -252932.163116, se: 780.599199411
Fix a typo in
parzen_ll.pyThe following is the result using my instruction on how to reproduce the original GAN and wish this would help others