def model(request):
mdl = cnn.NgnCnn(layer_size=10)
mdl.to(device)
def fin():
print("Teardown model")
request.addfinalizer(fin)
return mdl
def train_evaluation(parameters): # TODO
nfolds = 3
results = np.zeros(nfolds)
for i in range(nfolds):
net = cnn.NgnCnn(args.neurons, seed=i, excite=parameters["excite"])
net.to(DEVICE)
train(net, data_loader, datatype, DEVICE, parameters)
result = evaluate(net, data_loader, DEVICE)
results[i] = result["Loss"][0]
return {"cross_entropy": (results.mean(), results.std())}
},
}
results = pd.DataFrame(
index=range(REPEATS * len(group_config)),
columns=["Group", "Train Accuracy", "Test Accuracy", "Repeat"],
)
counter = 0
# DATASET
data_loader = cnn.Cifar10_data(mode="test", batch_size=BATCH_SIZE)
for i in range(REPEATS):
print(f"Running trial {i+1}/{REPEATS}")
net = cnn.NgnCnn(args.neurons, seed=i)
for group in group_config:
net_copy = copy.deepcopy(net)
if "Dropout" in group:
net_copy.dropout = 0.2
if "Neural Noise" in group:
net_copy.neural_noise = (-0.2, 0.5)
net_copy.to(device)
parameters = group_config[group]
log, _ = cnn.train_model(
model=net_copy,
dataset=data_loader,
dtype=dtype,
device=device,
**parameters,
)
results = pd.DataFrame(
index=range(REPEATS * len(group_config)),
columns=list(range(EPOCHS)) + ["Group", "Test Accuracy", "Repeat"],
)
counter = 0
mode = 'targetted' if args.targetted else 'random'
# DATASET
data_loader = cnn.Cifar10_data(mode="test", batch_size=BATCH_SIZE)
dfs = []
for i in range(REPEATS):
print(f"Running trial {i+1}/{REPEATS}")
net = cnn.NgnCnn(args.neurons, seed=i, excite=args.excite)
for group in group_config:
net_copy = copy.deepcopy(net)
net_copy.to(device)
parameters = group_config[group]
log = cnn.train_model(
model=net_copy,
dataset=data_loader,
dtype=dtype,
device=device,
**parameters,
)
# accuracy['Accuracy'][0]
accuracy = cnn.predict(net_copy, data_loader)['Accuracy'][0]
start = cnn.predict(net_copy, data_loader, train=True)['Accuracy'][0]
"epochs": EPOCHS,
"neurogenesis": 0,
"early_stop": False,
"optim_args":{
"lr": LR,
}
},
}
# DATASET
data_loader = cnn.Cifar10_data(mode="validation", data_folder=DATA_DIR, batch_size=BATCH_SIZE, num_workers=16)
REPEATS = 1
for i in range(REPEATS):
print(f"Running trial {i+1}/{REPEATS}")
net = cnn.NgnCnn(args.neurons, channels=3, seed=i, excite=True)
for group in group_config:
net_copy = copy.deepcopy(net)
net_copy.to(device)
if group == 'Dropout':
net_copy.dropout = 0.2
parameters = group_config[group]
log, optimizer = cnn.train_model(
model=net_copy,
dataset=data_loader,
dtype=dtype,
device=device,
**parameters,
)
accuracy = cnn.predict(net_copy, data_loader, device=device, valid=True)
},
}
# DATASET
data_loader = cnn.Cifar10_data(mode="validation",
data_folder=DATA_DIR,
batch_size=BATCH_SIZE,
num_workers=16)
REPEATS = 20
results = pd.DataFrame(index=range(REPEATS * len(group_config)),
columns=['Group', 'Test Accuracy', 'Repeat'])
counter = 0
for i in range(REPEATS):
print(f"Running trial {i+1}/{REPEATS}")
net = cnn.NgnCnn(args.neurons, channels=3, seed=i, neural_noise=(-0.2, 1))
for group in group_config:
net_copy = copy.deepcopy(net)
net_copy.to(device)
parameters = group_config[group]
log, optimizer = cnn.train_model(
model=net_copy,
dataset=data_loader,
dtype=dtype,
device=device,
**parameters,
)
accuracy = cnn.predict(net_copy,
data_loader,
device=device,
def train_evaluation(parameters, nfolds=5):
net = cnn.NgnCnn(args.neurons, seed)
net.to(DEVICE)
train(net, data_loader, datatype, DEVICE, parameters)
return evaluate(net, data_loader, DEVICE)