def experiment(args):
"""
Fits model to predict glossary terms based on textbook
chapter content.
Analyzes these models' efficacy on a test set, and reports
model performance.
"""
experiment_name = args["--expt"]
experiments.run_experiment(experiment_name, args)
def main(final_hyperparameters):
with open("./datasets/amazon_sparse.pkl", "rb") as f:
amazon_data = pickle.load(f)
train_X = amazon_data['Xtr'].toarray()
train_y = amazon_data['Ytr'].toarray()
test_X, test_y = None, None
if final_hyperparameters:
test_X = amazon_data['Xte'].toarray()
test_y = amazon_data['Yte'].toarray()
# Split dataset into training and validation
perm = np.random.RandomState(0).permutation(train_X.shape[0])
validation_X = train_X[perm[1201:], :]
validation_y = train_y[perm[1201:]]
train_X = train_X[perm[:1200], :]
train_y = train_y[perm[:1200]]
# You can try different seeds and check the model's performance!
seeds = np.random.RandomState(0).randint(low=0, high=65536, size=(10))
train_accuracies = []
validation_accuracies = []
test_accuracies = []
AMAZON_HYPERPARAMETERS["debug"] = False
AMAZON_HYPERPARAMETERS["num_classes"] = 50
for seed in seeds:
AMAZON_HYPERPARAMETERS["rng"] = np.random.RandomState(seed)
train_accuracy, validation_accuracy, test_accuracy = run_experiment(
AMAZON_HYPERPARAMETERS, train_X, train_y, validation_X,
validation_y, test_X, test_y)
print(
f"Seed: {seed} - Train Accuracy: {train_accuracy} - Validation Accuracy: {validation_accuracy} - Test Accuracy: {test_accuracy}"
)
train_accuracies.append(train_accuracy)
validation_accuracies.append(validation_accuracy)
test_accuracies.append(test_accuracy)
print(
f"Train Accuracies - Mean: {np.mean(train_accuracies)} - Standard Deviation: {np.std(train_accuracies, ddof=0)}"
)
print(
f"Validation Accuracies - Mean: {np.mean(validation_accuracies)} - Standard Deviation: {np.std(validation_accuracies, ddof=0)}"
)
print(
f"Test Accuracies - Mean: {np.mean(test_accuracies)} - Standard Deviation: {np.std(test_accuracies, ddof=0)}"
)
def main(final_hyperparameters):
with open("./datasets/occupancy.pkl", "rb") as f:
occupancy_data = pickle.load(f)
# Training Data
train_X = occupancy_data['Xtr']
train_y = occupancy_data['Ytr']
# Validation Data
validation_X = occupancy_data['Xte']
validation_y = occupancy_data['Yte']
# Testing Data
test_X, test_y = None, None
if final_hyperparameters:
test_X = occupancy_data['Xte2']
test_y = occupancy_data['Yte2']
# You can try different seeds and check the model's performance!
seeds = np.random.RandomState(0).randint(low=0, high=65536, size=(10))
train_accuracies = []
validation_accuracies = []
test_accuracies = []
OCCUPANCY_HYPERPARAMETERS["debug"] = False
OCCUPANCY_HYPERPARAMETERS["num_classes"] = 2
for seed in seeds:
OCCUPANCY_HYPERPARAMETERS["rng"] = np.random.RandomState(seed)
train_accuracy, validation_accuracy, test_accuracy = run_experiment(
OCCUPANCY_HYPERPARAMETERS, train_X, train_y, validation_X,
validation_y, test_X, test_y)
print(
f"Seed: {seed} - Train Accuracy: {train_accuracy} - Validation Accuracy: {validation_accuracy} - Test Accuracy: {test_accuracy}"
)
train_accuracies.append(train_accuracy)
validation_accuracies.append(validation_accuracy)
test_accuracies.append(test_accuracy)
print(
f"Train Accuracies - Mean: {np.mean(train_accuracies)} - Standard Deviation: {np.std(train_accuracies, ddof=0)}"
)
print(
f"Validation Accuracies - Mean: {np.mean(validation_accuracies)} - Standard Deviation: {np.std(validation_accuracies, ddof=0)}"
)
print(
f"Test Accuracies - Mean: {np.mean(test_accuracies)} - Standard Deviation: {np.std(test_accuracies, ddof=0)}"
)
"enable": 0.0, # float value other than 0.0 will enable
"world_size": 300.0, # don't set to very small value otherwise cloud movement looks discrete
"speed_multiplier": 5.0, # larger values give faster moving cloud (shadow)
"coverage_modifier": 0.0, # -1.0 ~ 1.0, larger value gives larger coverage of shadow
}),
"rotational_light": AttrDict({ # light rotates about x-axis within +-interval with fixed step size
"enable": 0.0, # float value other than 0.0 will enable
"interval": 10.0, # light rotate in the range of light_original_rotation +- interval
"step": 1.0 # larger number gives faster rotating light source
}),
"position": AttrDict({ # starting position of the object
"x": -120.9,
"y": 27.4834,
"z": 792.7
}),
"landing_zone": AttrDict({
"enable": 0.0, # float value other than 0.0 will enable
"offset": AttrDict({
"x": 31.5,
"y": -47.4,
"z": 23.5
})
})
})
####
env = make_unity_env(config)
run_experiment(env,
logger_kwargs={'base_dir': './results_ver2/Basic-Freefall'},
controller_kwargs={'disable_control': True})
"coverage_modifier": 0.0, # -1.0 ~ 1.0, larger value gives larger coverage of shadow
}),
"rotational_light": AttrDict({ # light rotates about x-axis within +-interval with fixed step size
"enable": 0.0, # float value other than 0.0 will enable
"interval": 10.0, # light rotate in the range of light_original_rotation +- interval
"step": 1.0 # larger number gives faster rotating light source
}),
"position": AttrDict({ # starting position of the object
"x": -120.9,
"y": 27.4834,
"z": 792.7
}),
"landing_zone": AttrDict({
"enable": 0.0, # float value other than 0.0 will enable
"offset": AttrDict({
"x": 31.5,
"y": -47.4,
"z": 23.5
})
})
})
####
env = make_unity_env(config)
run_experiment(env,
logger_kwargs={'base_dir': './results_ver2/Basic'},
controller_kwargs={
'max_velocity': 24,
'smooth_window': 10
})
"enable": 0.0, # float value other than 0.0 will enable
"world_size": 300.0, # don't set to very small value otherwise cloud movement looks discrete
"speed_multiplier": 5.0, # larger values give faster moving cloud (shadow)
"coverage_modifier": 0.0, # -1.0 ~ 1.0, larger value gives larger coverage of shadow
}),
"rotational_light": AttrDict({ # light rotates about x-axis within +-interval with fixed step size
"enable": 0.0, # float value other than 0.0 will enable
"interval": 10.0, # light rotate in the range of light_original_rotation +- interval
"step": 1.0 # larger number gives faster rotating light source
}),
"position": AttrDict({ # starting position of the object
"x": -120.9,
"y": 27.4834,
"z": 792.7
}),
"landing_zone": AttrDict({
"enable": 0.0, # float value other than 0.0 will enable
"offset": AttrDict({
"x": 31.5,
"y": -47.4,
"z": 23.5
})
})
})
####
env = make_unity_env(config)
run_experiment(env,
logger_kwargs={'base_dir': './results_ver2/Oracle'},
controller_kwargs={'max_velocity': 24, 'smooth_window': 10, 'use_oracle': True})
import os.path
import sys
import experiments
args = sys.argv
if len(args) > 1:
experiment_fname = args[1]
print(experiment_fname)
if os.path.exists(experiment_fname):
experiments.load_and_plot(experiment_fname)
else:
print(experiments.run_experiment(experiment_fname))
loss, global_step=global_step)
eval_metric_ops = None
if mode == tf.estimator.ModeKeys.EVAL:
eval_metric_ops = {
"accuracy": tf.metrics.accuracy(labels=labels,
predictions=predictions)
}
return tf.estimator.EstimatorSpec(mode,
predictions=predictions,
loss=loss,
train_op=train_op,
eval_metric_ops=eval_metric_ops)
hparams = {"batch_size": 128}
result, _ = experiments.run_experiment(
estimator_fn=my_dnn,
hparams=hparams,
train_data=train_data,
validation_data=val_data,
test_data=test_data,
model_dir="runs/test",
max_steps=15000,
#steps_before_early_stopping=8000,
early_stopping_fn=experiments.early_stopping("accuracy", operator.le))
print(result.results)
if mode == tf.estimator.ModeKeys.EVAL:
eval_metric_ops = {
"accuracy": tf.metrics.accuracy(labels=labels,
predictions=predictions)
}
return tf.estimator.EstimatorSpec(mode,
predictions=predictions,
loss=loss,
train_op=train_op,
eval_metric_ops=eval_metric_ops)
hparams = {
"batch_size": 128,
"activation": "relu",
"dropout": 0.5,
"optimizer": "Adam",
"learning_rate": 1e-4
}
result, _ = experiments.run_experiment(estimator_fn=my_cnn,
hparams=hparams,
train_data=train_data,
validation_data=val_data,
test_data=test_data,
model_dir="runs/test",
max_steps=5000)
print(result.results)
return ResNetPoolMixed(640512,
dim=(3 * 3**9, ),
n_channels=1,
batch_size=batch_size,
args=args)
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument("-d", "-data", help="gtzan, mtat or msd")
parser.add_argument("-logging", help="Logs to csv file")
parser.add_argument("-gpu", type=list, help="Run on gpu's, and which")
parser.add_argument("-local", help="Whether to run local or on server")
parser.add_argument("-cross",
help="Whether to run cross experiments or not")
args = parser.parse_args()
build_model = None
if args.local == 'True':
build_model = build_basic
else:
build_model = build_resnet_mixed_layer
#check_weights(build_model().build_model(), "C:\\Users\\kkr\\Desktop\\Thesis\\best_weights_max_average_net_1.6e-05.hdf5")
#check_weight(build_model().build_model(), "C:\\Users\\kkr\\Desktop\\Thesis\\best_weights_mixed_net_8e-05.hdf5")
if args.cross:
exp.run_cross_experiment(build_model, args)
else:
exp.run_experiment(build_model, args)