# Copyright (c) 2023, NVIDIA CORPORATION. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import Union
from nvflare.app_common.abstract.metric_comparator import MetricComparator
[docs]class Constant:
TN_PREFIX_CYCLIC = "cyclic"
TN_PREFIX_SWARM = "swarm"
TN_PREFIX_CROSS_SITE_EVAL = "cse"
BASENAME_CONFIG = "config"
BASENAME_START = "start"
BASENAME_LEARN = "learn"
BASENAME_EVAL = "eval"
BASENAME_REPORT_LEARN_RESULT = "report_learn_result"
BASENAME_REPORT_FINAL_RESULT = "report_final_result"
BASENAME_ASK_FOR_MODEL = "ask_for_model"
TASK_NAME_PREFIX = "cwf.task_prefix"
PRIVATE_P2P = "cwf.private_p2p"
ERROR = "cwf.error"
ORDER = "cwf.order"
CLIENTS = "cwf.clients"
START_CLIENT = "cwf.start_client"
RESULT_CLIENTS = "cwf.result_clients"
CLIENT_ORDER = "cwf.client_order"
LAST_ROUND = "cwf.last_round"
START_ROUND = "cwf.start_round"
TIMESTAMP = "cwf.timestamp"
ACTION = "cwf.action"
ALL_DONE = "cwf.all_done"
AGGR_CLIENTS = "cwf.aggr_clients"
TRAIN_CLIENTS = "cwf.train_clients"
AGGREGATOR = "cwf.aggr"
METRIC = "cwf.metric"
CLIENT = "cwf.client"
ROUND = "cwf.round"
CONFIG = "cwf.config"
STATUS_REPORTS = "cwf.status_reports"
RESULT = "cwf.result"
RESULT_TYPE = "cwf.result_type"
EVAL_LOCAL = "cwf.eval_local"
EVAL_GLOBAL = "cwf.eval_global"
EVALUATORS = "cwf.evaluators"
EVALUATEES = "cwf.evaluatees"
GLOBAL_CLIENT = "cwf.global_client"
MODEL_OWNER = "cwf.model_owner"
MODEL_NAME = "cwf.model_name"
MODEL_TYPE = "cwf.model_type"
GLOBAL_NAMES = "cwf.global_names"
EXECUTOR = "cwf.executor"
EXECUTOR_INITIALIZED = "cwf.executor_initialized"
EXECUTOR_FINALIZED = "cwf.executor_finalized"
TOPIC_SHARE_RESULT = "cwf.share_result"
TOPIC_END_WORKFLOW = "cwf.end_wf"
RC_NO_GLOBAL_MODELS = "cwf.no_global_models"
RC_NO_LOCAL_MODEL = "cwf.no_local_model"
RC_UNABLE_TO_EVAL = "cwf.unable_to_eval"
CONFIG_TASK_TIMEOUT = 300
START_TASK_TIMEOUT = 10
END_WORKFLOW_TIMEOUT = 2.0
TASK_CHECK_INTERVAL = 0.5
JOB_STATUS_CHECK_INTERVAL = 2.0
PER_CLIENT_STATUS_REPORT_TIMEOUT = 90.0
WORKFLOW_PROGRESS_TIMEOUT = 3600.0
LEARN_TASK_CHECK_INTERVAL = 1.0
LEARN_TASK_ACK_TIMEOUT = 10
LEARN_TASK_ABORT_TIMEOUT = 5.0
FINAL_RESULT_ACK_TIMEOUT = 10
GET_MODEL_TIMEOUT = 10
MAX_TASK_TIMEOUT = 3600
PROP_KEY_TRAIN_CLIENTS = "cwf.train_clients"
[docs]class ModelType:
LOCAL = "local"
GLOBAL = "global"
[docs]class ResultType:
BEST = "best"
LAST = "last"
[docs]class CyclicOrder:
FIXED = "fixed"
RANDOM = "random"
[docs]class StatusReport:
def __init__(
self,
timestamp=None,
action: str = "",
last_round=None,
all_done=False,
error: str = "",
):
self.timestamp = timestamp
self.action = action
self.last_round = last_round
self.all_done = all_done
self.error = error
[docs] def to_dict(self) -> dict:
result = {
Constant.TIMESTAMP: self.timestamp,
Constant.ACTION: self.action,
Constant.ALL_DONE: self.all_done,
}
if self.last_round is not None:
result[Constant.LAST_ROUND] = self.last_round
if self.error:
result[Constant.ERROR] = self.error
return result
def __eq__(self, other):
if not isinstance(other, StatusReport):
# don't attempt to compare against unrelated types
return ValueError(f"cannot compare to object of type {type(other)}")
return (
self.last_round == other.last_round
and self.timestamp == other.timestamp
and self.action == other.action
and self.all_done == other.all_done
and self.error == other.error
)
[docs]def status_report_from_dict(d: dict) -> StatusReport:
last_round = d.get(Constant.LAST_ROUND)
timestamp = d.get(Constant.TIMESTAMP)
all_done = d.get(Constant.ALL_DONE)
error = d.get(Constant.ERROR)
action = d.get(Constant.ACTION)
return StatusReport(
last_round=last_round,
timestamp=timestamp,
action=action,
all_done=all_done,
error=error,
)
[docs]def rotate_to_front(item, items: list):
num_items = len(items)
idx = items.index(item)
if idx != 0:
new_list = [None] * num_items
for i in range(num_items):
new_pos = i - idx
if new_pos < 0:
new_pos += num_items
new_list[new_pos] = items[i]
for i in range(num_items):
items[i] = new_list[i]
[docs]def topic_for_end_workflow(wf_id):
return f"{Constant.TOPIC_END_WORKFLOW}.{wf_id}"
[docs]def make_task_name(prefix: str, base_name: str) -> str:
return f"{prefix}_{base_name}"
[docs]class NumberMetricComparator(MetricComparator):
[docs] def compare(self, a, b) -> Union[int, float]:
if not isinstance(a, (int, float)):
raise ValueError(f"metric value must be a number but got {type(a)}")
if not isinstance(b, (int, float)):
raise ValueError(f"metric value must be a number but got {type(b)}")
return a - b