def analyze(): exp_params = ExperimentParameters() model_output_dir = "../../../models/qlearning/state/%i" % exp_params.seed exp_params.model_output_dir = model_output_dir image_output_dir = "../../../images/qlearning/state/%i" % exp_params.seed exp_params.image_output_dir = image_output_dir analyze_models(exp_params)
def analyze():
exp_params = ExperimentParameters()
model_output_dir = "../../../models/sarsa/reward/%i" % exp_params.seed
exp_params.model_output_dir = model_output_dir
image_output_dir = "../../../images/sarsa/reward/%i" % exp_params.seed
exp_params.image_output_dir = image_output_dir
analyze_models(exp_params)
def analyze():
exp_params = ExperimentParameters()
exp_params.seed = get_reward_seeds()[2]
model_output_dir = "../../../models/qlearning/reward/%i" % exp_params.seed
exp_params.model_output_dir = model_output_dir
image_output_dir = "../../../images/qlearning/reward/%i" % exp_params.seed
exp_params.image_output_dir = image_output_dir
analyze_models(exp_params)
def analyze(): exp_params = ExperimentParameters() exp_params.seed = get_parameters_seed() model_output_dir = "../../../models/expected_sarsa/params/%i" % exp_params.seed exp_params.model_output_dir = model_output_dir image_output_dir = "../../../images/expected_sarsa/params/%i" % exp_params.seed exp_params.image_output_dir = image_output_dir analyze_models(exp_params)
def train():
for params in get_snake_parameters():
env = SnakeEnvironment(params)
params.policy = EpsilonGreedyPolicy(env, params.epsilon)
exp_params = ExperimentParameters()
exp_params.env = env
exp_params.model_class = Sarsa
exp_params.model_params = params
exp_params.seed = get_parameters_seed()
output_dir = "../../../models/sarsa/params/%i" % exp_params.seed
exp_params.model_output_dir = output_dir
train_and_store_model(exp_params)
def analyze_test():
params = SnakeParameters()
params.discount_factor = StaticDiscountFactor(0.95)
params.learning_rate = StaticLearningRate(0.15)
env = SnakeEnvironment(params)
params.policy = GreedyPolicy(env)
exp_params = ExperimentParameters()
exp_params.env = env
exp_params.model_class = Sarsa
exp_params.model_params = params
model_output_dir = "../../../models/sarsa/reward/%i" % exp_params.seed
exp_params.model_output_dir = model_output_dir
image_output_dir = "../../../images/sarsa/reward/%i" % exp_params.seed
exp_params.image_output_dir = image_output_dir
analyze_models_test(exp_params)
def train():
params = SnakeParameters()
params.discount_factor = StaticDiscountFactor(0.95)
params.learning_rate = StaticLearningRate(0.15)
params.reward = get_state_reward()
env = SnakeEnvironment(params)
params.policy = EpsilonGreedyPolicy(env, params.epsilon)
exp_params = ExperimentParameters()
exp_params.env = env
exp_params.model_class = Sarsa
exp_params.model_params = params
for seed in get_state_seeds():
exp_params.seed = seed
output_dir = "../../../models/sarsa/state/%i" % exp_params.seed
exp_params.model_output_dir = output_dir
train_models(exp_params)
def train():
params = SnakeParameters()
params.discount_factor = StaticDiscountFactor(0.85)
params.learning_rate = StaticLearningRate(0.85)
env = SnakeEnvironment(params)
params.policy = EpsilonGreedyPolicy(env, params.epsilon)
exp_params = ExperimentParameters()
exp_params.env = env
exp_params.model_class = QLearning
exp_params.model_params = params
for seed in get_reward_seeds():
exp_params.seed = seed
output_dir = "../../../models/qlearning/reward/%i" % exp_params.seed
exp_params.model_output_dir = output_dir
for state in get_reward_states():
exp_params.model_params.state = state
train_models(exp_params)