Add --ef-runtime option to filter search experiments by ef values

Author: fcostaoliveira

engine/base_client/client.py

@@ -91,6 +91,7 @@ def run_experiment(
         upload_start_idx: int = 0,
         upload_end_idx: int = -1,
         num_queries: int = -1,
+        ef_runtime: List[int] = [],
     ):
         execution_params = self.configurator.execution_params(
             distance=dataset.config.distance, vector_size=dataset.config.vector_size
@@ -152,10 +153,18 @@ def run_experiment(
                 if "search_params" in search_params:
                     ef = search_params["search_params"].get("ef", "default")
                 client_count = search_params.get("parallel", 1)
+
+                # Filter by client count if parallels is specified
                 filter_client_count = len(parallels) > 0
                 if filter_client_count and (client_count not in parallels):
                     print(f"\tSkipping ef runtime: {ef}; #clients {client_count}")
                     continue
+
+                # Filter by ef runtime if ef_runtime is specified
+                filter_ef_runtime = len(ef_runtime) > 0
+                if filter_ef_runtime and isinstance(ef, int) and (ef not in ef_runtime):
+                    print(f"\tSkipping ef runtime: {ef}; #clients {client_count} (not in ef_runtime filter)")
+                    continue
                 for repetition in range(1, REPETITIONS + 1):
                     print(
                         f"\tRunning repetition {repetition} ef runtime: {ef}; #clients {client_count}"

run.py

@@ -27,6 +27,7 @@ def run(
     upload_start_idx: int = 0,
     upload_end_idx: int = -1,
     queries: int = typer.Option(-1, help="Number of queries to run. If the available queries are fewer, they will be reused."),
+    ef_runtime: List[int] = typer.Option([], help="Filter search experiments by ef runtime values. Only experiments with these ef values will be run."),
 ):
     """
     Example:
@@ -70,6 +71,7 @@ def run(
                         upload_start_idx,
                         upload_end_idx,
                         queries,
+                        ef_runtime,
                     )
                 client.delete_client()