def main(argv):
options = {
'batch_size': 128,
'max_time': 100.0,
'logdir': '/home/zenna/repos/inverse/log',
'template': template_dict,
'nnet_enhanced_pi': False,
'pointwise_pi': True,
'min_fx_y': False,
'nnet': True,
'min_fx_param': False,
'rightinv_pi_fx': False,
'nruns': 2
}
min_param_size = 10
param_types = {
'theta':
tensor_type(dtype=tf.float32,
shape=(options['batch_size'], min_param_size),
name="shrunk_param")
}
param_gen = {
k: infinite_samples(np.random.rand, v['shape'])
for k, v in param_types.items()
}
shrunk_param_gen = dictionary_gen(param_gen)
return compare(render_gen_graph, render_fwd_f, param_types,
shrunk_param_gen, options)
def main(argv):
global stats
options = {
'batch_size': 512,
'max_time': 5.0,
'logdir': '/home/zenna/repos/inverse/log',
'template': template_dict,
'nnet_enhanced_pi': True,
'pointwise_pi': True,
'min_fx_y': True,
'nnet': True
}
gen_graph = rand_gen_graph
fwd_f = rand_fwd_f
min_param_size = 1
param_types = {
'theta':
tensor_type(dtype=tf.float32,
shape=(options['batch_size'], min_param_size),
name="shrunk_param")
}
param_gen = {
k: infinite_samples(np.random.rand, v['shape'])
for k, v in param_types.items()
}
np.random.seed(0)
shrunk_param_gen = dictionary_gen(param_gen)
np.random.seed(0)
stats = compare(gen_graph, rand_fwd_f, param_types, shrunk_param_gen,
options)
return stats