from learner import Learner
from models import MlpPolicy, MlpValueFn
hparams = utils.Hyperparameters(
max_updates=50,
policy_hidden_dims=128,
value_fn_hidden_dims=128,
batch_size=32,
gamma=0.99,
rho_bar=1.0,
c_bar=1.0,
# policy_lr=1e-3,
# value_fn_lr=1e-3,
lr=1e-3,
policy_loss_c=1,
v_loss_c=0.5,
entropy_c=0.0006,
max_timesteps=1000,
queue_lim=8,
max_norm=10,
n_actors=1,
env_name="CartPole-v1", # "RacecarBulletEnv-v0",
log_path="./logs/",
save_every=50,
eval_every=2,
eval_eps=20,
verbose=1,
render=False,
)
if __name__ == "__main__":
metrics_string = " ; ".join("{}: {:05.3f}".format(k, v)
for k, v in metrics_mean.items())
logging.info("- Eval metrics : " + metrics_string)
return metrics_mean, loss
if __name__ == '__main__':
"""
Evaluate the model on the test set.
"""
# Load the parameters
args = parser.parse_args()
json_path = os.path.join(args.model_dir, 'params.json')
assert os.path.isfile(
json_path), "No json configuration file found at {}".format(json_path)
params = utils.Hyperparameters(json_path)
# use GPU if available
params.cuda = torch.cuda.is_available() # use GPU is available
# Set the random seed for reproducible experiments
torch.manual_seed(230)
if params.cuda: torch.cuda.manual_seed(230)
# Get the logger
utils.set_logger(os.path.join(args.model_dir, 'evaluate.log'))
# Create the input data pipeline
logging.info("Creating the dataset...")
# fetch dataloaders
json.dump(json_params, outfile)
if os.path.isfile(os.path.join(figures_dir, subdir + '.jpg')):
continue
cmd = "{python} train.py {} {} {} --bottleneck --model_dir={model_dir} --data_dir {data_dir}".format(
row['model'],
row['dataset'],
row['loss_fn'],
python=PYTHON,
model_dir=exp_dir,
data_dir='data')
print(cmd)
check_call(cmd, shell=True)
params = utils.Hyperparameters(os.path.join(exp_dir, 'params.json'))
params.cuda = torch.cuda.is_available()
torch.manual_seed(230)
if params.cuda:
torch.cuda.manual_seed(230)
utils.set_logger(os.path.join(exp_dir, 'lg.log'))
logging.info("Loading the datasets...")
# fetch dataloaders
test_dl = data_loader.fetch_test_dataloader(row['dataset'], 'data', params)
logging.info("- done.")
choices = {
'linear': LinearRegression().cuda() if params.cuda else LinearRegression(bottleneck=True),
'mlp': MLP(bottleneck=True).cuda() if params.cuda else MLP(bottleneck=True),
'cnn': CNN(bottleneck=True).cuda() if params.cuda else CNN(bottleneck=True)