text
stringlengths 1
93.6k
|
|---|
parser.add_argument('--num_repeat_pre', type=int,
|
help='Number of times the pre-training repeats',
|
default=2)
|
parser.add_argument('--num_repeat_train', type=int,
|
help='Number of times the training repeats',
|
default=15)
|
parser.add_argument('--seed', type=int, default=1,
|
help='Random seed')
|
parser.add_argument('--k_subgoal', type=int,
|
default=6, help='Number of actions to prune')
|
default_amr_server_ip = 'localhost'
|
parser.add_argument('--amr_server_ip', type=str,
|
default=default_amr_server_ip, help='IP for AMR server')
|
default_amr_server_port = 0
|
parser.add_argument('--amr_server_port', type=int,
|
default=default_amr_server_port,
|
help='Port number for AMR server')
|
args = parser.parse_args()
|
if args.amr_server_ip == default_amr_server_ip:
|
env_amr_server_ip = os.getenv('LOA_AMR_SERVER_IP', default_amr_server_ip)
|
else:
|
env_amr_server_ip = args.amr_server_ip
|
if args.amr_server_port == default_amr_server_port:
|
env_amr_server_port = int(os.getenv('LOA_AMR_SERVER_PORT',
|
str(default_amr_server_port)))
|
else:
|
env_amr_server_port = args.amr_server_port
|
print('AMR IP: %s, PORT: %s' % (env_amr_server_ip, env_amr_server_port))
|
filename = \
|
'loa-twc-dl%s-np%d-nt%d' % \
|
(args.difficulty_level, args.num_repeat_pre, args.num_repeat_train) + \
|
'-ks%d-sp%s' % (args.k_subgoal, args.sem_parser_mode)
|
results_folder = 'results/'
|
if not os.path.exists(results_folder):
|
os.mkdir(results_folder)
|
pkl_filepath = results_folder + filename + '.pkl'
|
np.random.seed(args.seed)
|
random.seed(args.seed)
|
torch.manual_seed(args.seed)
|
loa_agent = \
|
LOAAgent(
|
difficulty_level=args.difficulty_level,
|
amr_server_ip=env_amr_server_ip,
|
amr_server_port=env_amr_server_port,
|
admissible_verbs=None,
|
sem_parser_mode=args.sem_parser_mode,
|
num_repeats_pre=args.num_repeat_pre,
|
)
|
print('Admissible verbs: ', loa_agent.admissible_verbs)
|
loa_agent.extract_fact2logic(difficulty_level=args.difficulty_level,
|
repeats=args.num_repeat_train,
|
verbose=False, mincount=0.25)
|
starting_time = time.time()
|
loa_agent.reinforce_train_lnn(max_iters=1000,
|
verbose=False,
|
prune_low_rewards=True,
|
lam=0.0001)
|
print('Training time: %.2f' % (time.time() - starting_time))
|
print('Train eps:', loa_agent.train_eps)
|
print('Train steps:', loa_agent.steps)
|
loa_agent.save_pickel(pkl_filepath)
|
loa_agent = \
|
LOAAgent(
|
difficulty_level=args.difficulty_level,
|
amr_server_ip=env_amr_server_ip,
|
amr_server_port=env_amr_server_port,
|
admissible_verbs=None,
|
sem_parser_mode=args.sem_parser_mode
|
)
|
loa_agent.load_pickel(pkl_filepath)
|
print('Trained rules:')
|
loa_agent.display_rules()
|
perc_score, mean_steps = \
|
loa_agent.test_policy(difficulty_level=args.difficulty_level,
|
max_steps=50, split='test',
|
verbose=False, num_games=5)
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.