logger.info("Testing loss: {}".format(test_loss))
# Save the test scores
xp_manager.update_user_data(user_data={
"test_loss": test_loss,
})
tf.reset_default_graph()
with tf.Session(config=config_gpu) as sess_loop:
# compute the loop accuracy
logger.info("==================loop===================")
mode_to_evaluate = ["sampling", "greedy", "beam"]
cpu_pool = create_cpu_pool(args.no_thread, use_process=False)
train_batchifier = LooperBatchifier(tokenizer, generate_new_games=True)
eval_batchifier = LooperBatchifier(tokenizer, generate_new_games=False)
oracle_dir = "out/oracle/"
oracle_checkpoint = "6e0ec7f150b27f46296406853f498af6"
guesser_dir = "out/guesser/"
guesser_checkpoint = "c48036b430ebca1c44a25188edb05034"
oracle_config = get_config_from_xp(oracle_dir, oracle_checkpoint)
guesser_config = get_config_from_xp(guesser_dir, guesser_checkpoint)
qgen_network, qgen_batchifier_cstor = create_qgen(
config["model"], num_words=tokenizer.no_words)
qgen_var = [v for v in tf.global_variables()
if "qgen" in v.name] # and 'rl_baseline' not in v.name
for v in qgen_var:
print(v.name)
qgen_saver = tf.train.Saver(var_list=qgen_var)
# check that models are correctly loaded
test_model(sess, testset, cpu_pool=cpu_pool, tokenizer=tokenizer,
oracle=oracle_network,
guesser=guesser_network,
qgen=qgen_network,
batch_size=100,
logger=logger)
# create training tools
loop_sources = qgen_network.get_sources(sess)
logger.info("Sources: " + ', '.join(loop_sources))
evaluator = Evaluator(loop_sources, qgen_network.scope_name, network=qgen_network, tokenizer=tokenizer)
train_batchifier = LooperBatchifier(tokenizer, loop_sources, train=True)
eval_batchifier = LooperBatchifier(tokenizer, loop_sources, train=False)
# Initialize the looper to eval/train the game-simulation
qgen_network.build_sampling_graph(qgen_config["model"], tokenizer=tokenizer, max_length=loop_config['loop']['max_depth'])
looper_evaluator = BasicLooper(loop_config,
oracle=oracle_network,
guesser=guesser_network,
qgen=qgen_network,
tokenizer=tokenizer)
test_iterator = Iterator(testset, pool=cpu_pool,
batch_size=batch_size,
batchifier=eval_batchifier,
shuffle=False,
use_padding=True)
else:
qgen_wrapper = QGenUserWrapper(tokenizer)
logger.info("No QGen was registered >>> use user input")
looper_evaluator = BasicLooper(eval_config,
oracle_wrapper=oracle_wrapper,
guesser_wrapper=guesser_wrapper,
qgen_wrapper=qgen_wrapper,
tokenizer=tokenizer,
batch_size=1)
logs = []
# Start training
final_val_score = 0.
batchifier = LooperBatchifier(tokenizer, generate_new_games=False)
while True:
# Start new game
while True:
id_str = input(
'Do you want to play a new game? (Yes/No) --> ').lower()
if id_str == "y" or id_str == "yes": break
elif id_str == "n" or id_str == "no": exit(0)
# Pick id image
image_id = 0
while True:
id_str = int(
input(
'What is the image id you want to select? (-1 for random id) --> '