def __init__(self, predictor_id, config_info, request_q, reply_q,
predictor_name):
self.config_info = deepcopy(config_info)
self.predictor_id = predictor_id
self.request_q = request_q
self.reply_q = reply_q
self._report_period = 200
self._stats = PredictStats()
self.predictor_name = predictor_name
def __init__(self, thread_id, alg, request_q, reply_q, stats_deliver, lock):
self.alg = alg
self.thread_id = thread_id
self.request_q = request_q
self.reply_q = reply_q
self.lock = lock
self.stats_deliver = stats_deliver
self._report_period = 200
self._stats = PredictStats()
class PredictThread(object):
"""Predict Worker for async algorithm."""
def __init__(self, thread_id, alg, request_q, reply_q, stats_deliver,
lock):
self.alg = alg
self.thread_id = thread_id
self.request_q = request_q
self.reply_q = reply_q
self.lock = lock
self.stats_deliver = stats_deliver
self._report_period = 200
self._stats = PredictStats()
def predict(self):
"""Predict action."""
while True:
start_t0 = time()
data = self.request_q.recv()
state = get_msg_data(data)
self._stats.obs_wait_time += time() - start_t0
start_t1 = time()
with self.lock:
action = self.alg.predict(state)
self._stats.inference_time += time() - start_t1
set_msg_info(data, cmd="predict_reply")
set_msg_data(data, action)
# logging.debug("msg to explore: ", data)
self.reply_q.send(data)
self._stats.iters += 1
if self._stats.iters > self._report_period:
_report = self._stats.get()
self.stats_deliver.send(_report, block=True)
class Predictor(object):
"""Predict Worker for async algorithm."""
def __init__(self, predictor_id, config_info, request_q, reply_q,
predictor_name):
self.config_info = deepcopy(config_info)
self.predictor_id = predictor_id
self.request_q = request_q
self.reply_q = reply_q
self._report_period = 200
self._stats = PredictStats()
self.predictor_name = predictor_name
def process(self):
"""Predict action."""
while True:
start_t0 = time()
ctr_info, data = self.request_q.recv()
recv_data = {'ctr_info': ctr_info, 'data': data}
self._stats.obs_wait_time += time() - start_t0
cmd = ctr_info.get('sub_cmd', 'predict')
# if 'predict' in cmd:
# cmd = 'predict'
# else:
# print("sync model")
if cmd in self.process_fn.keys():
proc_fn = self.process_fn[cmd]
proc_fn(recv_data)
else:
raise KeyError("invalid cmd: {}".format(ctr_info['cmd']))
def sync_weights(self, recv_data):
model_weights = recv_data['data']
self.alg.set_weights(model_weights)
def predict(self, recv_data):
start_t1 = time()
state = get_msg_data(recv_data)
broker_id = get_msg_info(recv_data, 'broker_id')
explorer_id = get_msg_info(recv_data, 'explorer_id')
action = self.alg.predict(state)
self._stats.inference_time += time() - start_t1
reply_data = message(action,
cmd="predict_reply",
broker_id=broker_id,
explorer_id=explorer_id)
self.reply_q.put(reply_data)
self._stats.iters += 1
if self._stats.iters > self._report_period:
_report = self._stats.get()
reply_data = message(_report,
cmd="stats_msg{}".format(self.predictor_name))
self.reply_q.put(reply_data)
def start(self):
os.environ["CUDA_VISIBLE_DEVICES"] = str(-1)
alg_para = self.config_info.get('alg_para')
setproctitle.setproctitle("xt_predictor")
self.alg = alg_builder(**alg_para)
self.process_fn = {
'sync_weights': self.sync_weights,
'predict': self.predict
}
#start msg process
self.process()