Cheetah dev

python/fedml/__init__.py

@@ -25,7 +25,7 @@
 _global_training_type = None
 _global_comm_backend = None
 
-__version__ = "0.8.8a12"
+__version__ = "0.8.8a13"
 
 
 def init(args=None):

python/fedml/cli/README.md

@@ -157,41 +157,41 @@ At first, you need to define your job properties in the job yaml file, e.g. entr
 
 The job yaml file is as follows:
 ```
-fedml_params:
-    fedml_account_id: "111"
-    fedml_account_name: "fedml-demo"
-    project_name: Cheetah_HelloWorld
-    job_name: Cheetah_HelloWorld
+fedml_arguments:
+  fedml_account_id: "214"
+  fedml_account_name: "fedml-alex"
+  project_name: Cheetah_HelloWorld
+  job_name: Cheetah_HelloWorld080504
 
 # Local directory where your source code resides.
-work_dir: ~/falcon_examples
+Workspace: hello_world
 
 # Running entry commands which will be executed as the job entry point.
 # Support multiple lines, which can not be empty.
-run: | 
+Job: | 
     echo "Hello, Here is the Falcon platform."
     echo "Current directory is as follows."
     pwd
-    python train.py
+    python hello_world.py
 
 # Bootstrap shell commands which will be executed before running entry commands.
 # Support multiple lines, which can be empty.
-setup: |
-  pip install fedml              
+Bootstrap: |
+  pip install -r requirements.txt
   echo "Bootstrap finished."
-        
-gpu_requirements:
-    minimum_num_gpus: 1             # minimum # of GPUs to provision
-    maximum_cost_per_hour: $1.75    # max cost per hour for your job per machine
+
+computing:
+  minimum_num_gpus: 1             # minimum # of GPUs to provision
+  maximum_cost_per_hour: $1.75    # max cost per hour for your job per machine
 ```
 
 You just need to customize the following config items.
 
-1. `work_dir`, It is the local directory where your source code resides.
+1. `Workspace`, It is the local directory where your source code resides.
 
-2. `run`,  It is the running entry command which will be executed as the job entry point.
+2. `Job`,  It is the running entry command which will be executed as the job entry point.
 
-3. `setup`, It is the bootstrap shell command which will be executed before running entry commands.
+3. `Bootstrap`, It is the bootstrap shell command which will be executed before running entry commands.
 
 Then you can use the following example CLI to launch the job at the MLOps platform.
 (Replace $YourApiKey with your own account API key from open.fedml.ai)

python/fedml/cli/edge_deployment/client_runner.py

@@ -490,7 +490,7 @@ def execute_job_task(self, entry_file_full_path, conf_file_full_path, dynamic_ar
         if expert_mode is None:
             executable_interpreter = ClientConstants.CLIENT_SHELL_PS \
                 if platform.system() == ClientConstants.PLATFORM_WINDOWS else ClientConstants.CLIENT_SHELL_BASH
-            executable_commands = job_yaml.get("run", "")
+            executable_commands = job_yaml.get("Job", "")
         else:
             using_easy_mode = False
             executable_interpreter = expert_mode.get("executable_interpreter", "")
@@ -537,7 +537,7 @@ def execute_job_task(self, entry_file_full_path, conf_file_full_path, dynamic_ar
             logging.info("Run the client: {}".format(shell_cmd_list))
             process = ClientConstants.exec_console_with_shell_script_list(
                 shell_cmd_list,
-                should_capture_stdout=False,
+                should_capture_stdout=True,
                 should_capture_stderr=True
             )
             is_fl_task = False

python/fedml/cli/scheduler/README.md

@@ -13,41 +13,41 @@ At first, you need to define your job properties in the job yaml file, e.g. entr
 
 The job yaml file is as follows:
 ```
-fedml_params:
-    fedml_account_id: "111"
-    fedml_account_name: "fedml-demo"
-    project_name: Cheetah_HelloWorld
-    job_name: Cheetah_HelloWorld
-    
+fedml_arguments:
+  fedml_account_id: "214"
+  fedml_account_name: "fedml-alex"
+  project_name: Cheetah_HelloWorld
+  job_name: Cheetah_HelloWorld080504
+
 # Local directory where your source code resides.
-work_dir: ~/falcon_examples
+Workspace: hello_world
 
 # Running entry commands which will be executed as the job entry point.
 # Support multiple lines, which can not be empty.
-run: | 
+Job: | 
     echo "Hello, Here is the Falcon platform."
     echo "Current directory is as follows."
     pwd
-    python train.py
+    python hello_world.py
 
 # Bootstrap shell commands which will be executed before running entry commands.
 # Support multiple lines, which can be empty.
-setup: |
-  pip install fedml              
+Bootstrap: |
+  pip install -r requirements.txt
   echo "Bootstrap finished."
-        
-gpu_requirements:
-    minimum_num_gpus: 1             # minimum # of GPUs to provision
-    maximum_cost_per_hour: $1.75    # max cost per hour for your job per machine
+
+computing:
+  minimum_num_gpus: 1             # minimum # of GPUs to provision
+  maximum_cost_per_hour: $1.75    # max cost per hour for your job per machine
 ```
 
 You just need to customize the following config items. 
 
-1. `work_dir`, It is the local directory where your source code resides.
+1. `Workspace`, It is the local directory where your source code resides.
 
-2. `run`,  It is the running entry command which will be executed as the job entry point.
+2. `Job`,  It is the running entry command which will be executed as the job entry point.
 
-3. `setup`, It is the bootstrap shell command which will be executed before running entry commands.
+3. `Bootstrap`, It is the bootstrap shell command which will be executed before running entry commands.
 
 Then you can use the following example CLI to launch the job at the MLOps platform.
 (Replace $YourApiKey with your own account API key from open.fedml.ai)

python/fedml/cli/scheduler/call_gpu.yaml

@@ -1,26 +1,26 @@
-fedml_params:
+fedml_arguments:
   fedml_account_id: "214"
   fedml_account_name: "fedml-alex"
   project_name: Cheetah_HelloWorld
   job_name: Cheetah_HelloWorld
 
 # Local directory where your source code resides.
-work_dir: hello_world
+Workspace: hello_world
 
 # Running entry commands which will be executed as the job entry point.
 # Support multiple lines, which can not be empty.
-run: | 
+Job: | 
     echo "Hello, Here is the Falcon platform."
     echo "Current directory is as follows."
     pwd
-    python vision_transformer.py
+    python hello_world.py
 
 # Bootstrap shell commands which will be executed before running entry commands.
 # Support multiple lines, which can be empty.
-setup: |
+Bootstrap: |
   pip install -r requirements.txt
   echo "Bootstrap finished."
 
-gpu_requirements:
+computing:
   minimum_num_gpus: 1             # minimum # of GPUs to provision
   maximum_cost_per_hour: $1.75    # max cost per hour for your job per machine

python/fedml/cli/scheduler/hello_world/hello_world.py

@@ -1,90 +1,3 @@
-# import torch
-#
-import logging
-import time
-
-from fedml import FedMLRunner, mlops, constants
-import fedml
-
-# from fedml.data.MNIST.data_loader import download_mnist, load_partition_data_mnist
-#
-#
-# def load_data(args):
-#     download_mnist(args.data_cache_dir)
-#     fedml.logging.info("load_data. dataset_name = %s" % args.dataset)
-#
-#     """
-#     Please read through the data loader at to see how to customize the dataset for FedML framework.
-#     """
-#     (
-#         client_num,
-#         train_data_num,
-#         test_data_num,
-#         train_data_global,
-#         test_data_global,
-#         train_data_local_num_dict,
-#         train_data_local_dict,
-#         test_data_local_dict,
-#         class_num,
-#     ) = load_partition_data_mnist(
-#         args,
-#         args.batch_size,
-#         train_path=args.data_cache_dir + "/MNIST/train",
-#         test_path=args.data_cache_dir + "/MNIST/test",
-#     )
-#     """
-#     For shallow NN or linear models,
-#     we uniformly sample a fraction of clients each round (as the original FedAvg paper)
-#     """
-#     args.client_num_in_total = client_num
-#     dataset = [
-#         train_data_num,
-#         test_data_num,
-#         train_data_global,
-#         test_data_global,
-#         train_data_local_num_dict,
-#         train_data_local_dict,
-#         test_data_local_dict,
-#         class_num,
-#     ]
-#     return dataset, class_num
-#
-#
-# class LogisticRegression(torch.nn.Module):
-#     def __init__(self, input_dim, output_dim):
-#         super(LogisticRegression, self).__init__()
-#         self.linear = torch.nn.Linear(input_dim, output_dim)
-#
-#     def forward(self, x):
-#         import torch
-#         outputs = torch.sigmoid(self.linear(x))
-#         return outputs
-
 
 if __name__ == "__main__":
-
-    # Init logs before the program starts to log.
-    mlops.log_print_init()
-
-    # Use print or logging.info to print your logs, which will be uploaded to MLOps and can be showed in the logs page.
-    print("Hello world. Here is the Falcon platform.")
-    # logging.info("Hello world. Here is the Falcon platform.")
-
-    time.sleep(10)
-
-    # Cleanup logs when the program will be ended.
-    mlops.log_print_cleanup()
-
-    #
-    # # init device
-    # device = fedml.device.get_device(args)
-    #
-    # # load data
-    # dataset, output_dim = load_data(args)
-    #
-    # # load model (the size of MNIST image is 28 x 28)
-    # model = LogisticRegression(28 * 28, output_dim)
-    #
-    # # start training
-    # fedml_runner = FedMLRunner(args, device, dataset, model)
-    # fedml_runner.run()
+    print("Hi everyone, I am an Falcon job.")

python/fedml/cli/scheduler/launch_manager.py

@@ -383,16 +383,16 @@ def build_mlops_package(
 class FedMLJobConfig(object):
     def __init__(self, job_yaml_file):
         self.job_config_dict = load_yaml_config(job_yaml_file)
-        self.account_id = self.job_config_dict["fedml_params"]["fedml_account_id"]
-        self.account_name = self.job_config_dict["fedml_params"]["fedml_account_name"]
-        self.project_name = self.job_config_dict["fedml_params"]["project_name"]
-        self.job_name = self.job_config_dict["fedml_params"]["job_name"]
+        self.account_id = self.job_config_dict["fedml_arguments"]["fedml_account_id"]
+        self.account_name = self.job_config_dict["fedml_arguments"]["fedml_account_name"]
+        self.project_name = self.job_config_dict["fedml_arguments"]["project_name"]
+        self.job_name = self.job_config_dict["fedml_arguments"]["job_name"]
         self.base_dir = os.path.dirname(job_yaml_file)
         self.using_easy_mode = True
         self.executable_interpreter = "bash"
-        self.executable_file_folder = self.job_config_dict.get("work_dir", None)
-        self.executable_commands = self.job_config_dict.get("run", "")
-        self.bootstrap = self.job_config_dict.get("setup", None)
+        self.executable_file_folder = self.job_config_dict.get("Workspace", None)
+        self.executable_commands = self.job_config_dict.get("Job", "")
+        self.bootstrap = self.job_config_dict.get("Bootstrap", None)
         self.executable_file = None
         self.executable_conf_option = ""
         self.executable_conf_file_folder = None
@@ -441,8 +441,8 @@ def __init__(self, job_yaml_file):
         self.executable_file = str(self.executable_file).replace('\\', os.sep).replace('/', os.sep)
         self.executable_conf_file = str(self.executable_conf_file).replace('\\', os.sep).replace('/', os.sep)
 
-        self.minimum_num_gpus = self.job_config_dict["gpu_requirements"]["minimum_num_gpus"]
-        self.maximum_cost_per_hour = self.job_config_dict["gpu_requirements"]["maximum_cost_per_hour"]
+        self.minimum_num_gpus = self.job_config_dict["computing"]["minimum_num_gpus"]
+        self.maximum_cost_per_hour = self.job_config_dict["computing"]["maximum_cost_per_hour"]
         self.application_name = FedMLJobConfig.generate_application_name(self.job_name, self.project_name)
 
     @staticmethod

python/setup.py

@@ -91,7 +91,7 @@ def finalize_options(self):
 
 setup(
     name="fedml",
-    version="0.8.8a12",
+    version="0.8.8a13",
     author="FedML Team",
     author_email="[email protected]",
     description="A research and production integrated edge-cloud library for "