def sample_binomial(p):
"""Samples elementwise from the binomial distribution with
probability p"""
if use_debug_rng:
r = myrand.rand(p.shape)
else:
r = gp.rand(p.shape)
# n = np.random.random(p.shape)
# n = gp.rand(p.shape)
# r = gp.zeros(p.shape)
return r < p
n_cols = 784
n_iters = 60000 / n_rows * 3
def write_bytes_to_file(file, data):
for x in data:
file.write(chr(int(x)))
# use gnumpy rng
gp.seed_rand(1)
with open("rng_gnumpy.dat", "wb") as file:
for i in range(n_iters):
print "%d / %d\r" % (i, n_iters),
gx = gp.rand((n_rows, n_cols))
x = gp.as_numpy_array(gx)
fx = np.reshape(x, -1)
bx = np.floor(fx * 256)
write_bytes_to_file(file, bx)
# use lcg rng
mr.seed(1)
with open("rng_lcg.dat", "wb") as file:
for i in range(n_iters):
print "%d / %d\r" % (i, n_iters),
gx = mr.rand((n_rows, n_cols))
x = gp.as_numpy_array(gx)
fx = np.reshape(x, -1)
bx = np.floor(fx * 256)
write_bytes_to_file(file, bx)
X, TX = rbmutil.load_mnist(False)
# load ruslan's training set
mdata = scipy.io.loadmat("mnist.mat")
X = gp.as_garray(mdata['fbatchdata'])
# create output directory
rbmutil.enter_rbm_plot_directory("mnist", cfg.n_hid, cfg.use_pcd,
cfg.n_gibbs_steps, "training_from.txt",
clean=False)
# Build RBM
rbm = RestrictedBoltzmannMachine(cfg.batch_size, cfg.n_vis, cfg.n_hid,
cfg.n_gibbs_steps)
mr.seed(30)
rbm.weights = 0.01 * (0.5 - mr.rand(rbm.weights.shape))
# load weights
if use_ruslans_start_weights:
filename = "matlab_epoch%d.mat" % (from_epoch-1 + 1)
print "Loading Ruslan's start weights from %s" % filename
mdata = scipy.io.loadmat(filename)
ml.rbm.weights = gp.as_garray(mdata['vishid'])
ml.rbm.bias_vis = gp.as_garray(mdata['visbiases'])
ml.rbm.bias_hid = gp.as_garray(mdata['hidbiases'])
else:
print "Loading python start weights"
filename = "weights-%02d.npz" % (from_epoch-1)
rbmutil.load_parameters(rbm, filename)
# initialize momentums