Allow feature_shape to be passed to fx plan initialize

openfl/interface/cli.py

@@ -254,7 +254,7 @@ def entry():
         cli.add_command(command_group.__getattribute__(module))
 
     try:
-        cli()
+        cli(max_content_width=120)
     except Exception as e:
         error_handler(e)
 

openfl/interface/plan.py

@@ -12,6 +12,8 @@
 from click import Path as ClickPath
 
 from openfl.utilities.path_check import is_directory_traversal
+from openfl.utilities.click_types import InputSpec
+from openfl.utilities.mocks import MockDataLoader
 
 logger = getLogger(__name__)
 
@@ -36,12 +38,15 @@ def plan(context):
         default='plan/data.yaml', type=ClickPath(exists=True))
 @option('-a', '--aggregator_address', required=False,
         help='The FQDN of the federation agregator')
-@option('-f', '--feature_shape', required=False,
-        help='The input shape to the model')
+@option('-f', '--input_shape', cls=InputSpec, required=False,
+        help="The input shape to the model. May be provided as a list:\n\n"
+             "--input_shape [1,28,28]\n\n"
+             "or as a dictionary for multihead models (must be passed in quotes):\n\n"
+             "--input_shape \"{'input_0': [1, 240, 240, 4],'output_1': [1, 240, 240, 1]}\"\n\n ")
 @option('-g', '--gandlf_config', required=False,
         help='GaNDLF Configuration File Path')
 def initialize(context, plan_config, cols_config, data_config,
-               aggregator_address, feature_shape, gandlf_config):
+               aggregator_address, input_shape, gandlf_config):
     """
     Initialize Data Science plan.
 
@@ -73,18 +78,15 @@ def initialize(context, plan_config, cols_config, data_config,
 
     init_state_path = plan.config['aggregator']['settings']['init_state_path']
 
-    # TODO:  Is this part really needed?  Why would we need to collaborator
-    #  name to know the input shape to the model?
-
-    # if  feature_shape is None:
-    #     if  cols_config is None:
-    #         exit('You must specify either a feature
-    #         shape or authorized collaborator
-    #         list in order for the script to determine the input layer shape')
-
-    collaborator_cname = list(plan.cols_data_paths)[0]
-
-    data_loader = plan.get_data_loader(collaborator_cname)
+    # This is needed to bypass data being locally available
+    if input_shape is not None:
+        logger.info('Attempting to generate initial model weights with'
+                    f' custom input shape {input_shape}')
+        data_loader = MockDataLoader(input_shape)
+    else:
+        # If feature shape is not provided, data is assumed to be present
+        collaborator_cname = list(plan.cols_data_paths)[0]
+        data_loader = plan.get_data_loader(collaborator_cname)
     task_runner = plan.get_task_runner(data_loader)
     tensor_pipe = plan.get_tensor_pipe()
 

openfl/utilities/click_types.py

@@ -1,8 +1,9 @@
-# Copyright (C) 2020-2023 Intel Corporation
+# Copyright (C) 2020-2024 Intel Corporation
 # SPDX-License-Identifier: Apache-2.0
-"""Click types module."""
+"""Custom input types definition for Click"""
 
 import click
+import ast
 
 from openfl.utilities import utils
 
@@ -31,5 +32,17 @@ def convert(self, value, param, ctx):
         return value
 
 
+class InputSpec(click.Option):
+    """List or dictionary that corresponds to the input shape for a model"""
+    def type_cast_value(self, ctx, value):
+        try:
+            if value is None:
+                return None
+            else:
+                return ast.literal_eval(value)
+        except Exception:
+            raise click.BadParameter(value)
+
+
 FQDN = FqdnParamType()
 IP_ADDRESS = IpAddressParamType()

openfl/utilities/mocks.py

@@ -0,0 +1,12 @@
+# Copyright (C) 2020-2024 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+"""Mock objects to eliminate extraneous dependencies"""
+
+
+class MockDataLoader:
+    """Placeholder dataloader for when data is not available"""
+    def __init__(self, feature_shape):
+        self.feature_shape = feature_shape
+
+    def get_feature_shape(self):
+        return self.feature_shape