Dex launcher dev

examples/advanced/loose-integration/data-exchange/h5/av_job/app_cn/config/config_fed_client.json

@@ -0,0 +1,30 @@
+{
+  "format_version": 2,
+
+  "executors": [
+    {
+      "tasks": ["train"],
+      "executor": {
+        "path": "custom_executor.CustomExecutor",
+        "args": {
+          "data_exchanger_id": "data_exchanger",
+          "data_exchange_path": "/tmp/nvflare/av_cn"
+        }
+      }
+    }
+  ],
+  "task_result_filters": [
+  ],
+  "task_data_filters": [
+  ],
+  "components": [
+    {
+      "id": "data_exchanger",
+      "path": "nvflare.app_opt.h5.data_exchanger.H5DataExchanger",
+      "args": {
+        "pipe_role": "x",
+        "heartbeat_timeout": 0
+      }
+    }
+  ]
+}

examples/advanced/loose-integration/data-exchange/h5/av_job/app_cn/custom/custom_executor.py

@@ -0,0 +1,64 @@
+# 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 Optional
+
+from nvflare.apis.fl_context import FLContext
+from nvflare.apis.shareable import Shareable
+from nvflare.app_common.executors.launcher_executor import (
+    LauncherExecutor,
+    model_data_to_shareable,
+    shareable_to_model_data,
+)
+
+
+class CustomExecutor(LauncherExecutor):
+    def __init__(
+        self,
+        data_exchanger_id: str,
+        data_exchange_path: str,
+        file_accessor_id: Optional[str] = None,
+        get_timeout: Optional[float] = 10000,
+    ):
+        """CustomExecutor for autonomous vehicle training."""
+        super().__init__(
+            data_exchanger_id=data_exchanger_id,
+            file_accessor_id=file_accessor_id,
+            data_exchange_path=data_exchange_path,
+        )
+
+        self._timeout = get_timeout
+        self._rounds = 0
+
+    def _prepare_for_launch(self, shareable: Shareable):
+        # dump weights for outer script to read
+        model_data = shareable_to_model_data(shareable=shareable)
+        self.data_exchanger.put(self._from_nvflare, data=model_data)
+        self.data_exchanger.put("round_starts", {"round": self._rounds})
+
+    def _get_result(self, task_name: str, fl_ctx: FLContext) -> Shareable:
+        try:
+            output_model_data = self.data_exchanger.get(self._to_nvflare, self._timeout)
+            round_ends = self.data_exchanger.get("round_ends", self._timeout)
+            external_round = round_ends["round"]
+            if external_round != self._rounds:
+                raise RuntimeError("rounds mismatch.")
+            self._rounds += 1
+            result = model_data_to_shareable(output_model_data)
+            return result
+        except Exception as e:
+            err_msg = f"External training is not finished within timeout ({self._timeout}) seconds: {e}."
+            self.log_exception(fl_ctx, err_msg)
+            self.system_panic(err_msg, fl_ctx)
+            return None

examples/advanced/loose-integration/data-exchange/h5/av_job/app_server/config/config_fed_server.json

@@ -0,0 +1,54 @@
+{
+  "format_version": 2,
+
+  "server": {
+    "heart_beat_timeout": 600
+  },
+  "task_data_filters": [],
+  "task_result_filters": [],
+  "components": [
+    {
+      "id": "persistor",
+      "name": "PTFileModelPersistor",
+      "args": {
+        "model": {
+          "path": "simple_network.SimpleNetwork"
+        }
+      }
+    },
+    {
+      "id": "shareable_generator",
+      "path": "nvflare.app_common.shareablegenerators.full_model_shareable_generator.FullModelShareableGenerator",
+      "args": {}
+    },
+    {
+      "id": "aggregator",
+      "path": "nvflare.app_common.aggregators.intime_accumulate_model_aggregator.InTimeAccumulateWeightedAggregator",
+      "args": {
+        "expected_data_kind": "WEIGHTS"
+      }
+    },
+    {
+      "id": "tb_analytics_receiver",
+      "name": "TBAnalyticsReceiver",
+      "args": {"events": ["fed.analytix_log_stats"]}
+    }
+  ],
+  "workflows": [
+      {
+        "id": "scatter_and_gather",
+        "name": "ScatterAndGather",
+        "args": {
+            "min_clients" : 2,
+            "num_rounds" : 10,
+            "start_round": 0,
+            "wait_time_after_min_received": 0,
+            "aggregator_id": "aggregator",
+            "persistor_id": "persistor",
+            "shareable_generator_id": "shareable_generator",
+            "train_task_name": "train",
+            "train_timeout": 0
+        }
+      }
+  ]
+}

examples/advanced/loose-integration/data-exchange/h5/av_job/app_server/custom/simple_network.py

@@ -0,0 +1,38 @@
+# Copyright (c) 2021, NVIDIA CORPORATION.
+#
+# 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.
+
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+
+class SimpleNetwork(nn.Module):
+    def __init__(self):
+        super(SimpleNetwork, self).__init__()
+
+        self.conv1 = nn.Conv2d(3, 6, 5)
+        self.pool = nn.MaxPool2d(2, 2)
+        self.conv2 = nn.Conv2d(6, 16, 5)
+        self.fc1 = nn.Linear(16 * 5 * 5, 120)
+        self.fc2 = nn.Linear(120, 84)
+        self.fc3 = nn.Linear(84, 10)
+
+    def forward(self, x):
+        x = self.pool(F.relu(self.conv1(x)))
+        x = self.pool(F.relu(self.conv2(x)))
+        x = torch.flatten(x, 1)  # flatten all dimensions except batch
+        x = F.relu(self.fc1(x))
+        x = F.relu(self.fc2(x))
+        x = self.fc3(x)
+        return x

examples/advanced/loose-integration/data-exchange/h5/av_job/app_us/config/config_fed_client.json

@@ -0,0 +1,30 @@
+{
+  "format_version": 2,
+
+  "executors": [
+    {
+      "tasks": ["train"],
+      "executor": {
+        "path": "custom_executor.CustomExecutor",
+        "args": {
+          "data_exchanger_id": "data_exchanger",
+          "data_exchange_path": "/tmp/nvflare/av_us"
+        }
+      }
+    }
+  ],
+  "task_result_filters": [
+  ],
+  "task_data_filters": [
+  ],
+  "components": [
+    {
+      "id": "data_exchanger",
+      "path": "nvflare.app_opt.h5.data_exchanger.H5DataExchanger",
+      "args": {
+        "pipe_role": "x",
+        "heartbeat_timeout": 0
+      }
+    }
+  ]
+}

examples/advanced/loose-integration/data-exchange/h5/av_job/app_us/custom/custom_executor.py

@@ -0,0 +1,64 @@
+# 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 Optional
+
+from nvflare.apis.fl_context import FLContext
+from nvflare.apis.shareable import Shareable
+from nvflare.app_common.executors.launcher_executor import (
+    LauncherExecutor,
+    model_data_to_shareable,
+    shareable_to_model_data,
+)
+
+
+class CustomExecutor(LauncherExecutor):
+    def __init__(
+        self,
+        data_exchanger_id: str,
+        data_exchange_path: str,
+        file_accessor_id: Optional[str] = None,
+        get_timeout: Optional[float] = 10000,
+    ):
+        """CustomExecutor for autonomous vehicle training."""
+        super().__init__(
+            data_exchanger_id=data_exchanger_id,
+            file_accessor_id=file_accessor_id,
+            data_exchange_path=data_exchange_path,
+        )
+
+        self._timeout = get_timeout
+        self._rounds = 0
+
+    def _prepare_for_launch(self, shareable: Shareable):
+        # dump weights for outer script to read
+        model_data = shareable_to_model_data(shareable=shareable)
+        self.data_exchanger.put(self._from_nvflare, data=model_data)
+        self.data_exchanger.put("round_starts", {"round": self._rounds})
+
+    def _get_result(self, task_name: str, fl_ctx: FLContext) -> Shareable:
+        try:
+            output_model_data = self.data_exchanger.get(self._to_nvflare, self._timeout)
+            round_ends = self.data_exchanger.get("round_ends", self._timeout)
+            external_round = round_ends["round"]
+            if external_round != self._rounds:
+                raise RuntimeError("rounds mismatch.")
+            self._rounds += 1
+            result = model_data_to_shareable(output_model_data)
+            return result
+        except Exception as e:
+            err_msg = f"External training is not finished within timeout ({self._timeout}) seconds: {e}."
+            self.log_exception(fl_ctx, err_msg)
+            self.system_panic(err_msg, fl_ctx)
+            return None

examples/advanced/loose-integration/data-exchange/h5/av_job/meta.json

@@ -0,0 +1,16 @@
+{
+  "name": "av",
+  "resource_spec": {},
+  "min_clients" : 2,
+  "deploy_map": {
+    "app_server": [
+      "server"
+    ],
+    "app_us": [
+      "site-1"
+    ],
+    "app_cn": [
+      "site-2"
+    ]
+  }
+}