Merge pull request #830 from FedML-AI/dev/v0.7.0

update to v0.8.0

README.md

@@ -243,7 +243,7 @@ Here `hierarchical` means that inside each FL Client (data silo), there are mult
 
 ## **FedML Beehive Examples**
 
-- [Federated Learning on Android Smartphones](./doc/en/cross-device/examples/mqtt_s3_fedavg_mnist_lr_example.md)
+- [Federated Learning on Android Smartphones](./doc/en/cross-device/examples/cross_device_android_example.md)
 
 
 # FedML on Smartphone and IoTs

android/fedmlsdk/src/main/java/ai/fedml/edge/service/ClientAgentManager.java

@@ -75,7 +75,7 @@ public void registerMessageReceiveHandlers(final long edgeId) {
         edgeCommunicator.subscribe(startTrainTopic, (OnTrainStartListener) this::handleTrainStart);
         final String stopTrainTopic = "flserver_agent/" + edgeId + "/stop_train";
         edgeCommunicator.subscribe(stopTrainTopic, (OnTrainStopListener) this::handleTrainStop);
-        final String MLOpsQueryStatusTopic = "/mlops/report_device_status";
+        final String MLOpsQueryStatusTopic = "mlops/report_device_status";
         edgeCommunicator.subscribe(MLOpsQueryStatusTopic, (OnMLOpsMsgListener) this::handleMLOpsMsg);
 
         final String exitTrainWithExceptionTopic = "flserver_agent/" + edgeId + "/exit_train_with_exception";

android/fedmlsdk/src/main/java/ai/fedml/edge/service/communicator/message/MessageDefine.java

@@ -44,8 +44,8 @@ public interface MessageDefine {
 
 
     // Client Status
-    String MQTT_LAST_WILL_TOPIC = "/flclient_agent/last_will_msg";
-    String MQTT_REPORT_ACTIVE_STATUS_TOPIC = "/flclient_agent/active";
+    String MQTT_LAST_WILL_TOPIC = "flclient_agent/last_will_msg";
+    String MQTT_REPORT_ACTIVE_STATUS_TOPIC = "flclient_agent/active";
 
     String MSG_MLOPS_CLIENT_STATUS_OFFLINE = "OFFLINE";
     String MSG_MLOPS_CLIENT_STATUS_IDLE = "IDLE";

android/fedmlsdk/src/main/java/ai/fedml/edge/service/component/ProfilerEventLogger.java

@@ -7,7 +7,7 @@
 import ai.fedml.edge.utils.LogHelper;
 
 public class ProfilerEventLogger {
-    private static final String EVENT_TOPIC = "/mlops/events";
+    private static final String EVENT_TOPIC = "mlops/events";
     private static final int EVENT_TYPE_STARTED = 0;
     private static final int EVENT_TYPE_ENDED = 1;
 

devops/k8s/README_MODEL_SERVING.md

@@ -4,7 +4,7 @@ This tutorial will guide you to deploy your models to target computing devices,
 
 The entire workflow is as follows:
 1. create a model card by uploading your trained model file and related configuration (YAML)
-2. bind (login) computing resource to FedML MLOps model serving platform (https://model.fedml.ai)
+2. bind (login) computing resource to FedML MLOps model serving platform (https://open.fedml.ai)
    - Kubernetes mode
    - CLI mode
 3. start the deployment and get the inference API once the deployment is finished
@@ -14,7 +14,7 @@ When your model deployment is finished, you will get an endpoint URL and inferen
 
 ```curl -XPOST https://$YourEndPointIngressDomainName/inference/api/v1/predict -H 'accept: application/json' -d'{  "model_version": "v11-Thu Jan 05 08:20:24 GMT 2023",  "model_name": "model_340_18_fedml_test_model_v11-Thu-Jan-05-08-20-24-GMT-2023",  "data": "This is our test data. Please fill in here with your real data.",  "end_point_id": 336,  "model_id": 18,  "token": "2e081ef115d04ee8adaffe5c1d0bfbac"}'```
 
- You may run your model deployment flow via the ModelOps(model.fedml.ai) and CLI.
+ You may run your model deployment flow via the ModelOps(open.fedml.ai) and CLI.
 
 Model Deployment CLI:
 
@@ -23,7 +23,7 @@ fedml model deploy -n $model_name --on_premise -d $device_id_list -u $user_id -k
 
 e.g. fedml model deploy -n fedml_sample_model -u 1420 -k c9356b9c4ce44363bb66366b290201 -dt md.on_premise_device -d [178077,178076]
 
-Note: You may find your device id in the Computing Resource page at the ModelOps(model.fedml.ai) platform.
+Note: You may find your device id in the Computing Resource page at the ModelOps(open.fedml.ai) platform.
       In the $device_id_list, the master device should be the first item.
 ```
 
@@ -62,15 +62,15 @@ Inference end point ingress will be used as your model serving endpoint URL whic
    ```kubectl get nodes --show-labels```
 
 ### 4). Prepare parameters will be used in the next step.
- You should fetch $YourAccountId and $YourApiKey from ModelOps(model.fedml.ai) which will be used in the next step. 
+ You should fetch $YourAccountId and $YourApiKey from ModelOps(open.fedml.ai) which will be used in the next step. 
 
 ### 5). You may run the Helm Charts Installation commands to install FedML model serving packages to the above labeled nodes.
 
 ```kubectl create namespace $YourNameSpace```
 
-```helm install --set env.fedmlAccountId="$YourAccountId" --set env.fedmlApiKey="$YourApiKey" --set env.fedmlVersion="release"  fedml-model-premise-slave fedml-model-premise-slave-0.7.397.tgz -n $YourNameSpace```
+```helm install --set env.fedmlAccountId="$YourAccountId" --set env.fedmlApiKey="$YourApiKey" --set env.fedmlVersion="release"  fedml-model-premise-slave fedml-model-premise-slave-latest.tgz -n $YourNameSpace```
 
-```helm install --set env.fedmlAccountId="$YourAccountId" --set env.fedmlApiKey="$YourApiKey" --set env.fedmlVersion="release" --set "inferenceGateway.ingress.host=$YourEndPointIngressDomainName" --set "inferenceGateway.ingress.className=nginx" fedml-model-premise-master fedml-model-premise-master-0.7.397.tgz -n $YourNameSpace```
+```helm install --set env.fedmlAccountId="$YourAccountId" --set env.fedmlApiKey="$YourApiKey" --set env.fedmlVersion="release" --set "inferenceGateway.ingress.host=$YourEndPointIngressDomainName" --set "inferenceGateway.ingress.className=nginx" fedml-model-premise-master fedml-model-premise-master-latest.tgz -n $YourNameSpace```
 
 Notes: $YourEndPointIngressDomainName is your model serving end point URL host which will be used in your inference API, e.g.
 
@@ -137,7 +137,7 @@ List model in the remote model repository:
 Build local model repository as zip model package:
 ```fedml model package -n $model_name```
 
-Push local model repository to ModelOps(model.fedml.ai):
+Push local model repository to ModelOps(open.fedml.ai):
 ```fedml model push -n $model_name -u $user_id -k $user_api_key```
 
 Pull remote model(ModelOps) to local model repository:
@@ -158,4 +158,4 @@ A: Yes.
 
 
 4. Q: During deployment, what if the k8s service does not have a public IP? \
-A: During deployment, we don't need to initiate access to your k8s service from model.fedml.ai, only your k8s cluster can initiate access to model.fedml.ai
+A: During deployment, we don't need to initiate access to your k8s service from open.fedml.ai, only your k8s cluster can initiate access to open.fedml.ai

devops/k8s/fedml-model-inference-ingress/Chart.yaml

@@ -1,6 +1,6 @@
 apiVersion: v2
 name: fedml-model-inference-ingress
-description: A Helm chart for master on-premise device on FedML model serving platform(model.fedml.ai)
+description: A Helm chart for master on-premise device on FedML model serving platform(open.fedml.ai)
 
 # A chart can be either an 'application' or a 'library' chart.
 #
@@ -15,7 +15,7 @@ type: application
 # This is the chart version. This version number should be incremented each time you make changes
 # to the chart and its templates, including the app version.
 # Versions are expected to follow Semantic Versioning (https://semver.org/)
-version: 0.7.377
+version: 0.7.700
 
 # This is the version number of the application being deployed. This version number should be
 # incremented each time you make changes to the application. Versions are not expected to

devops/k8s/fedml-model-inference-ingress/values.yaml

@@ -43,7 +43,7 @@ service:
 ingress:
   enabled: true
   className: ""
-  annotations: {}
+  annotations:
     kubernetes.io/ingress.class: nginx
     ingress.kubernetes.io/cors-allow-headers: '*'
     ingress.kubernetes.io/cors-allow-methods: 'PUT, GET, POST, OPTIONS, HEAD, DELETE, PATCH'
@@ -59,7 +59,7 @@ ingress:
     nginx.ingress.kubernetes.io/ssl-redirect: 'true'
     # kubernetes.io/tls-acme: "true"
   hosts:
-    - host: model.fedml.ai
+    - host: open.fedml.ai
       paths:
         - path: /inference
           pathType: Prefix

devops/k8s/fedml-model-premise-master/Chart.yaml

@@ -1,6 +1,6 @@
 apiVersion: v2
 name: fedml-model-premise-master
-description: A Helm chart for master on-premise device on FedML model serving platform(model.fedml.ai)
+description: A Helm chart for master on-premise device on FedML model serving platform(open.fedml.ai)
 
 # A chart can be either an 'application' or a 'library' chart.
 #
@@ -15,7 +15,7 @@ type: application
 # This is the chart version. This version number should be incremented each time you make changes
 # to the chart and its templates, including the app version.
 # Versions are expected to follow Semantic Versioning (https://semver.org/)
-version: 0.7.397
+version: 0.7.700
 
 # This is the version number of the application being deployed. This version number should be
 # incremented each time you make changes to the application. Versions are not expected to

devops/k8s/fedml-model-premise-master/values.yaml

@@ -58,7 +58,7 @@ ingress:
     nginx.ingress.kubernetes.io/ssl-redirect: 'true'
     # kubernetes.io/tls-acme: "true"
   hosts:
-    - host: model.fedml.ai
+    - host: open.fedml.ai
       paths:
         - path: /
           pathType: Prefix
@@ -146,7 +146,7 @@ inferenceGateway:
       nginx.ingress.kubernetes.io/proxy-send-timeout: '6000'
       nginx.ingress.kubernetes.io/ssl-redirect: 'true'
       # kubernetes.io/tls-acme: "true"
-    host: model.fedml.ai
+    host: open.fedml.ai
     protocol: https #http
     tls: [ ]
     #  - secretName: chart-example-tls

devops/k8s/fedml-model-premise-slave/Chart.yaml

@@ -1,6 +1,6 @@
 apiVersion: v2
 name: fedml-model-premise-slave
-description: A Helm chart for slave on-premise device on FedML model serving platform(model.fedml.ai)
+description: A Helm chart for slave on-premise device on FedML model serving platform(open.fedml.ai)
 
 # A chart can be either an 'application' or a 'library' chart.
 #
@@ -15,7 +15,7 @@ type: application
 # This is the chart version. This version number should be incremented each time you make changes
 # to the chart and its templates, including the app version.
 # Versions are expected to follow Semantic Versioning (https://semver.org/)
-version: 0.7.397
+version: 0.7.700
 
 # This is the version number of the application being deployed. This version number should be
 # incremented each time you make changes to the application. Versions are not expected to

python/examples/cross_silo/mqtt_s3_fedavg_mnist_lr_example/custom_data_and_model/config/fedml_config.yaml

@@ -23,7 +23,7 @@ train_args:
   client_id_list:
   client_num_in_total: 1
   client_num_per_round: 1
-  comm_round: 10
+  comm_round: 2
   epochs: 1
   batch_size: 10
   client_optimizer: sgd

python/examples/cross_silo/mqtt_s3_fedavg_mnist_lr_example/custom_data_and_model/torch_server.py

@@ -52,7 +52,6 @@ def __init__(self, input_dim, output_dim):
         self.linear = torch.nn.Linear(input_dim, output_dim)
 
     def forward(self, x):
-        import torch
         outputs = torch.sigmoid(self.linear(x))
         return outputs
 

python/fedml/__init__.py

@@ -23,7 +23,7 @@
 _global_training_type = None
 _global_comm_backend = None
 
-__version__ = "0.7.601"
+__version__ = "0.8.0"
 
 
 def init(args=None):

python/fedml/cli/cli.py

@@ -767,7 +767,7 @@ def device():
     pass
 
 
-@device.command("login", help="Login as model device agent(MDA) on the ModelOps platform (model.fedml.ai).")
+@device.command("login", help="Login as model device agent(MDA) on the ModelOps platform (open.fedml.ai).")
 @click.argument("userid", nargs=-1)
 @click.option(
     "--cloud", "-c", default=None, is_flag=True, help="login as fedml cloud device.",
@@ -834,7 +834,7 @@ def login_as_model_device_agent(
                                                    redis_addr, redis_port, redis_password)
 
 
-@device.command("logout", help="Logout from the ModelOps platform (model.fedml.ai)")
+@device.command("logout", help="Logout from the ModelOps platform (open.fedml.ai)")
 @click.option(
     "--slave", "-s", default=None, is_flag=True, help="logout from slave device.",
 )
@@ -932,12 +932,19 @@ def list_models(name):
     default="release",
     help="interact with which version of ModelOps platform. It should be dev, test or release",
 )
-def list_remote_models(name, user, api_key, version):
+@click.option(
+    "--local_server",
+    "-ls",
+    type=str,
+    default="127.0.0.1",
+    help="local server address.",
+)
+def list_remote_models(name, user, api_key, version, local_server):
     if user is None or api_key is None:
         click.echo("You must provide arguments for User Id and Api Key (use -u and -k options).")
         return
     FedMLModelCards.get_instance().set_config_version(version)
-    model_query_result = FedMLModelCards.get_instance().list_models(name, user, api_key)
+    model_query_result = FedMLModelCards.get_instance().list_models(name, user, api_key, local_server)
     if model_query_result is None or model_query_result.model_list is None or len(model_query_result.model_list) <= 0:
         click.echo("Model list is empty.")
     else:
@@ -962,7 +969,7 @@ def package_model(name):
         click.echo("Failed to build model {}.".format(name))
 
 
-@model.command("push", help="Push local model repository to ModelOps(model.fedml.ai).")
+@model.command("push", help="Push local model repository to ModelOps(open.fedml.ai).")
 @click.option(
     "--name", "-n", type=str, help="model name.",
 )
@@ -985,15 +992,23 @@ def package_model(name):
     default="release",
     help="interact with which version of ModelOps platform. It should be dev, test or release",
 )
-def push_model(name, model_storage_url, model_net_url, user, api_key, version):
+@click.option(
+    "--local_server",
+    "-ls",
+    type=str,
+    default="127.0.0.1",
+    help="local server address.",
+)
+def push_model(name, model_storage_url, model_net_url, user, api_key, version, local_server):
     if user is None or api_key is None:
         click.echo("You must provide arguments for User Id and Api Key (use -u and -k options).")
         return
     FedMLModelCards.get_instance().set_config_version(version)
     model_is_from_open = True if model_storage_url is not None and model_storage_url != "" else False
     model_storage_url, model_zip = FedMLModelCards.get_instance().push_model(name, user, api_key,
                                                                              model_storage_url=model_storage_url,
-                                                                             model_net_url=model_net_url)
+                                                                             model_net_url=model_net_url,
+                                                                             local_server=local_server)
     if model_is_from_open:
         click.echo("Push model {} with model storage url {} successfully.".format(name, model_storage_url))
     else:
@@ -1022,19 +1037,26 @@ def push_model(name, model_storage_url, model_net_url, user, api_key, version):
     default="release",
     help="interact with which version of ModelOps platform. It should be dev, test or release",
 )
-def pull_model(name, user, api_key, version):
+@click.option(
+    "--local_server",
+    "-ls",
+    type=str,
+    default="127.0.0.1",
+    help="local server address.",
+)
+def pull_model(name, user, api_key, version, local_server):
     if user is None or api_key is None:
         click.echo("You must provide arguments for User Id and Api Key (use -u and -k options).")
         return
     FedMLModelCards.get_instance().set_config_version(version)
-    if FedMLModelCards.get_instance().pull_model(name, user, api_key):
+    if FedMLModelCards.get_instance().pull_model(name, user, api_key, local_server):
         click.echo("Pull model {} successfully.".format(name))
     else:
         click.echo("Failed to pull model {}.".format(name))
 
 
 @model.command("deploy",
-               help="Deploy specific model to ModelOps platform(model.fedml.ai) or just for local debugging deployment.")
+               help="Deploy specific model to ModelOps platform(open.fedml.ai) or just for local debugging deployment.")
 @click.option(
     "--name", "-n", type=str, help="model name.",
 )
@@ -1063,11 +1085,19 @@ def pull_model(name, user, api_key, version):
     default="release",
     help="interact with which version of ModelOps platform. It should be dev, test or release",
 )
+@click.option(
+    "--local_server",
+    "-ls",
+    type=str,
+    default="127.0.0.1",
+    help="local server address.",
+)
 @click.option(
     "--use_local_deployment", "-ld", default=None, is_flag=True,
     help="deploy local model repository by sending MQTT message(just use for debugging).",
 )
-def deploy_model(name, on_premise, cloud, devices, user, api_key, params, version, use_local_deployment):
+def deploy_model(name, on_premise, cloud, devices, user, api_key, params, version,
+                 local_server, use_local_deployment):
     if user is None or api_key is None:
         click.echo("You must provide arguments for User Id and Api Key (use -u and -k options).")
         return
@@ -1079,14 +1109,23 @@ def deploy_model(name, on_premise, cloud, devices, user, api_key, params, versio
     if is_cloud and is_on_premise:
         is_cloud = False
 
+    is_local_dev = use_local_deployment
+    if use_local_deployment is None:
+        is_local_dev = False
+
     if is_on_premise:
         device_type = "md.on_premise_device"
     else:
         device_type = "md.fedml_cloud_device"
     FedMLModelCards.get_instance().set_config_version(version)
-    paramsDict = json.loads(params) # load config from Cli
+
+    params_dict = {}
+    if is_local_dev:
+        params_dict = json.loads(params)  # load config from Cli
+
     if FedMLModelCards.get_instance().deploy_model(name, device_type, devices, user, api_key,
-                                                   paramsDict, use_local_deployment):
+                                                   params_dict, use_local_deployment,
+                                                   local_server):
         click.echo("Deploy model {} successfully.".format(name))
     else:
         click.echo("Failed to deploy model {}.".format(name))
@@ -1100,7 +1139,7 @@ def inference():
     pass
 
 
-@inference.command("query", help="Query inference parameters for specific model from ModelOps platform(model.fedml.ai).")
+@inference.command("query", help="Query inference parameters for specific model from ModelOps platform(open.fedml.ai).")
 @click.option(
     "--name", "-n", type=str, help="model name.",
 )
@@ -1115,7 +1154,7 @@ def query_model_infer(name):
         click.echo("Failed to query model {}.".format(name))
 
 
-@inference.command("run", help="Run inference action for specific model from ModelOps platform(model.fedml.ai).")
+@inference.command("run", help="Run inference action for specific model from ModelOps platform(open.fedml.ai).")
 @click.option(
     "--name", "-n", type=str, help="model name.",
 )