def mli_get_current_weights(self) -> Weights:
request = empty_pb2.Empty()
try:
response = self.stub.GetCurrentWeights(request)
weights = iterator_to_weights(response, decode=False)
return weights
except grpc.RpcError as ex:
_logger.exception(f"Failed to get_current_weights: {ex}")
raise Exception(f"Failed to get_current_weights: {ex}")
def mli_propose_weights(self) -> Weights:
request = empty_pb2.Empty()
try:
response = self.stub.ProposeWeights(request)
weights = iterator_to_weights(response, decode=False)
return weights
except grpc.RpcError as ex:
_logger.exception(f"Failed to train_model: {ex}")
raise ConnectionError(f"GRPC error: {ex}")
def test_iterator_and_back():
part_a = bytes(b"a" * WEIGHTS_PART_SIZE_BYTES)
part_b = bytes(b"b" * (WEIGHTS_PART_SIZE_BYTES - 2))
test_weights = part_a + part_b
weights = Weights(weights=test_weights)
iterator = weights_to_iterator(input_weights=weights, encode=False)
result = iterator_to_weights(request_iterator=iterator, decode=False)
assert result == weights
def test_in_order_iterator_to_weights():
test_weights = b"abc"
parts = [
WeightsPart(weights=test_weights[i:i + 1],
byte_index=i,
total_bytes=len(test_weights))
for i in range(len(test_weights))
]
result = iterator_to_weights(request_iterator=iter(parts), decode=False)
assert result.weights == test_weights
def SetWeights(self, request_iterator, context):
_count_set.inc()
if not self._check_model(context):
return empty_pb2.Empty()
self._learner_mutex.acquire()
try:
weights = iterator_to_weights(request_iterator)
self.learner.mli_accept_weights(weights)
except Exception as ex: # pylint: disable=W0703
_logger.exception(f"Exception in SetWeights: {ex} {type(ex)}")
context.set_code(grpc.StatusCode.INTERNAL)
context.set_details(str(ex))
_count_set_err.inc()
finally:
self._learner_mutex.release()
return empty_pb2.Empty()
def TestWeights(self, request_iterator, context):
_count_test.inc()
pw = ipb2.ProposedWeights()
if not self._check_model(context):
return pw
self._learner_mutex.acquire()
try:
_logger.debug("Test weights...")
weights = iterator_to_weights(request_iterator)
proposed_weights = self.learner.mli_test_weights(weights)
pw.vote_score = proposed_weights.vote_score
pw.test_score = proposed_weights.test_score
pw.vote = proposed_weights.vote
_logger.debug("Testing done!")
except Exception as ex: # pylint: disable=W0703
_logger.exception(f"Exception in TestWeights: {ex} {type(ex)}")
context.set_code(grpc.StatusCode.INTERNAL)
context.set_details(str(ex))
_count_test_err.inc()
finally:
self._learner_mutex.release()
return pw
