def test_loading_parsed_model(self, _model_1, _config):
model = _model_1
path_wd = _config.get('paths', 'path_wd')
model_name = _config.get('paths', 'filename_ann')
keras.models.save_model(model, os.path.join(path_wd,
model_name + '.h5'))
# Only perform parsing step.
updates = {
'tools': {
'evaluate_ann': False,
'normalize': False,
'convert': False,
'simulate': False
}
}
_config.read_dict(updates)
run_pipeline(_config)
path_parsed = os.path.join(
path_wd,
_config.get('paths', 'filename_parsed_model') + '.h5')
model_parsed = keras.models.load_model(path_parsed)
x = np.random.random_sample((5, ) + model.input_shape[1:])
y = model.predict(x)
y_parsed = model_parsed.predict(x)
assert np.allclose(y, y_parsed)
def test_maxpool_first(self, _model_maxpool2D_1_first, _config_first):
"""Test that maxpooling fails with data format channels_first."""
config = _config_first
path_wd = config.get('paths', 'path_wd')
model_name = config.get('paths', 'filename_ann')
models.save_model(_model_maxpool2D_1_first,
os.path.join(path_wd, model_name + '.h5'))
updates = {
'tools': {
'evaluate_ann': False,
'normalize': False
},
'simulation': {
'duration': 100,
'num_to_test': 100,
'batch_size': 50
},
'output': {
'log_vars': {'activations_n_b_l', 'spiketrains_n_b_l_t'}
}
}
config.read_dict(updates)
run_pipeline(config)
def test_inisim(self, _model_2, _config):
path_wd = _config.get('paths', 'path_wd')
model_name = _config.get('paths', 'filename_ann')
keras.models.save_model(_model_2,
os.path.join(path_wd, model_name + '.h5'))
updates = {
'tools': {
'evaluate_ann': False
},
'simulation': {
'simulator': 'INI',
'duration': 100,
'num_to_test': 100,
'batch_size': 50,
'keras_backend': 'tensorflow'
},
'output': {
'log_vars': {'activations_n_b_l', 'spiketrains_n_b_l_t'}
}
}
_config.read_dict(updates)
acc = run_pipeline(_config)
assert acc[0] >= 0.95
corr = get_correlations(_config)
assert np.all(corr[:-1] > 0.99)
assert corr[-1] > 0.90
def test_spinnaker_sparse(self, _model_3, _config):
path_wd = _config.get('paths', 'path_wd')
model_name = _config.get('paths', 'filename_ann')
models.save_model(_model_3, os.path.join(path_wd, model_name + '.h5'))
updates = {
'tools': {
'evaluate_ann': False
},
'input': {
'poisson_input': True
},
'simulation': {
'simulator': 'spiNNaker',
'duration': 100,
'num_to_test': 1, # smaller to make more feasible
'batch_size': 1
},
'output': {
'log_vars': {'activations_n_b_l', 'spiketrains_n_b_l_t'}
}
}
_config.read_dict(updates)
initialize_simulator(_config)
acc = run_pipeline(_config)
assert acc[0] >= 0.95
corr = get_correlations(_config)
assert np.all(corr[:-1] > 0.97)
assert corr[-1] > 0.5
def test_maxpool_fallback(self, _model_maxpool2D_1, _config):
"""Test that maxpooling falls back on average pooling."""
path_wd = _config.get('paths', 'path_wd')
model_name = _config.get('paths', 'filename_ann')
models.save_model(_model_maxpool2D_1,
os.path.join(path_wd, model_name + '.h5'))
updates = {
'tools': {
'evaluate_ann': False,
'normalize': False
},
'conversion': {
'max2avg_pool': True
},
'simulation': {
'duration': 100,
'num_to_test': 100,
'batch_size': 50
},
'output': {
'log_vars': {'activations_n_b_l', 'spiketrains_n_b_l_t'}
}
}
_config.read_dict(updates)
acc = run_pipeline(_config)
assert acc[0] >= 0.8
corr = get_correlations(_config)
assert np.all(corr[:-1] > 0.99)
assert corr[-1] > 0.90
def test_maxpool(self, _model_maxpool2D_1_last, _config_last):
"""Test maxpooling."""
config = _config_last
path_wd = config.get('paths', 'path_wd')
model_name = config.get('paths', 'filename_ann')
models.save_model(_model_maxpool2D_1_last,
os.path.join(path_wd, model_name + '.h5'))
updates = {
'tools': {
'evaluate_ann': False,
'normalize': False
},
'simulation': {
'duration': 100,
'num_to_test': 100,
'batch_size': 50
},
'output': {
'log_vars': {'activations_n_b_l', 'spiketrains_n_b_l_t'}
}
}
config.read_dict(updates)
acc = run_pipeline(config)
acc_ann = get_ann_acc(config)
assert acc[0] >= 0.9 * acc_ann
corr = get_correlations(config)
assert np.all(corr[:-1] > 0.99)
assert corr[-1] > 0.90
def test_nest(self, _model_1, _config):
path_wd = _config.get('paths', 'path_wd')
model_name = _config.get('paths', 'filename_ann')
keras.models.save_model(_model_1,
os.path.join(path_wd, model_name + '.h5'))
updates = {
'tools': {'evaluate_ann': False},
'simulation': {
'simulator': 'nest',
'duration': 50,
'num_to_test': 10,
'batch_size': 1,
'dt': 0.1},
'cell': {
'tau_refrac': 0.1,
'delay': 0.1,
'v_thresh': 0.01},
'output': {
'log_vars': {'activations_n_b_l', 'spiketrains_n_b_l_t'}}}
_config.read_dict(updates)
initialize_simulator(_config)
acc = run_pipeline(_config)
assert acc[0] >= 0.95
corr = get_correlations(_config)
assert np.all(corr[:-1] > 0.97)
assert corr[-1] > 0.5
def test_parsing(self, _model_4_first):
self.prepare_model(_model_4_first)
updates = {
'tools': {
'evaluate_ann': True,
'parse': True,
'normalize': False,
'convert': False,
'simulate': False
},
'input': {
'model_lib': 'pytorch'
},
'simulation': {
'num_to_test': 100,
'batch_size': 50
}
}
self.config.read_dict(updates)
initialize_simulator(self.config)
acc = run_pipeline(self.config)
assert acc[0] >= 0.8
def test_pipeline(self, _model_4_first):
self.prepare_model(_model_4_first)
updates = {
'tools': {
'evaluate_ann': False
},
'input': {
'model_lib': 'pytorch'
},
'simulation': {
'duration': 100,
'num_to_test': 100,
'batch_size': 50
},
'output': {
'log_vars': {'activations_n_b_l', 'spiketrains_n_b_l_t'}
}
}
self.config.read_dict(updates)
initialize_simulator(self.config)
acc = run_pipeline(self.config)
assert acc[0] >= 0.8
corr = get_correlations(self.config)
assert np.all(corr[:-1] > 0.97)
assert corr[-1] > 0.5
def main(filepath=None):
"""Entry point for running the toolbox.
Note
----
There is no need to call this function directly, because python sets up an
executable during :ref:`installation` that can be called from terminal.
"""
from snntoolbox.bin.utils import update_setup, run_pipeline
if filepath is not None:
config = update_setup(filepath)
run_pipeline(config)
return
parser = argparse.ArgumentParser(
description='Run SNN toolbox to convert an analog neural network into '
'a spiking neural network, and optionally simulate it.')
parser.add_argument('config_filepath',
nargs='?',
help='Path to configuration file.')
parser.add_argument('-t',
'--terminal',
action='store_true',
help='Set this flag to run the toolbox from terminal. '
'Omit this flag to open GUI.')
args = parser.parse_args()
_filepath = os.path.abspath(args.config_filepath)
if _filepath is not None:
config = update_setup(_filepath)
if args.terminal:
run_pipeline(config)
else:
from snntoolbox.bin.gui import gui
gui.main()
else:
if args.terminal:
parser.error("When using the SNN toolbox from terminal, a "
"config_filepath argument must be provided.")
return
else:
from snntoolbox.bin.gui import gui
gui.main()
def test_loihi(self, _model_1, _config):
path_wd = _config.get('paths', 'path_wd')
model_name = _config.get('paths', 'filename_ann')
keras.models.save_model(_model_1,
os.path.join(path_wd, model_name + '.h5'))
updates = {
'tools': {
'evaluate_ann': True,
'normalize': False
},
'loihi': {
'reset_mode': 'soft',
'desired_threshold_to_input_ratio': 1,
'compartment_kwargs': {
'biasExp': 6,
'vThMant': 512
},
'connection_kwargs': {
'numWeightBits': 8,
'weightExponent': 0,
'numBiasBits': 12
},
'validate_partitions': False,
'save_output': False,
'do_overflow_estimate': False,
'normalize_thresholds': True
},
'simulation': {
'simulator': 'loihi',
'duration': 512,
'num_to_test': 100,
'batch_size': 1
},
'output': {
'log_vars': {'activations_n_b_l', 'spiketrains_n_b_l_t'}
}
}
_config.read_dict(updates)
initialize_simulator(_config)
acc = run_pipeline(_config)
assert acc[0] >= 0.95
corr = get_correlations(_config)
assert np.all(corr[:-1] > 0.97)
assert corr[-1] > 0.5
def test_nest(self, _model_1, _config):
path_wd = _config.get('paths', 'path_wd')
model_name = _config.get('paths', 'filename_ann')
keras.models.save_model(_model_1,
os.path.join(path_wd, model_name + '.h5'))
updates = {
'tools': {
'evaluate_ann': False
},
'input': {
'poisson_input': True
},
'simulation': {
'simulator': 'nest',
'duration': 100,
'num_to_test': 100,
'batch_size': 1
},
'output': {
'log_vars': {'activations_n_b_l', 'spiketrains_n_b_l_t'}
}
}
_config.read_dict(updates)
try:
initialize_simulator(_config)
except (ImportError, KeyError, ValueError):
return
acc = run_pipeline(_config)
assert acc[0] >= 0.95
corr = get_correlations(_config)
assert np.all(corr[:-1] > 0.97)
assert corr[-1] > 0.5
