aiagents4pharma 1.43.0__py3-none-any.whl → 1.45.0__py3-none-any.whl

aiagents4pharma/talk2knowledgegraphs/tools/milvus_multimodal_subgraph_extraction.py

@@ -2,27 +2,25 @@
 Tool for performing multimodal subgraph extraction.
 """
 
-# import datetime
-from typing import Type, Annotated
 import logging
+from typing import Annotated
+
 import hydra
 import pandas as pd
-from pydantic import BaseModel, Field
-from langchain_core.tools import BaseTool
 from langchain_core.messages import ToolMessage
+from langchain_core.tools import BaseTool
 from langchain_core.tools.base import InjectedToolCallId
-from langgraph.types import Command
 from langgraph.prebuilt import InjectedState
+from langgraph.types import Command
+from pydantic import BaseModel, Field
 from pymilvus import Collection
-from ..utils.extractions.milvus_multimodal_pcst import MultimodalPCSTPruning
+
+from ..utils.extractions.milvus_multimodal_pcst import (
+    DynamicLibraryLoader,
+    MultimodalPCSTPruning,
+    SystemDetector,
+)
 from .load_arguments import ArgumentData
-try:
-    import cupy as py
-    import cudf
-    df = cudf
-except ImportError:
-    import numpy as py
-    df = pd
 
 # Initialize logger
 logging.basicConfig(level=logging.INFO)
@@ -41,14 +39,10 @@ class MultimodalSubgraphExtractionInput(BaseModel):
         arg_data: Argument for analytical process over graph data.
     """
 
-    tool_call_id: Annotated[str, InjectedToolCallId] = Field(
-        description="Tool call ID."
-    )
+    tool_call_id: Annotated[str, InjectedToolCallId] = Field(description="Tool call ID.")
     state: Annotated[dict, InjectedState] = Field(description="Injected state.")
     prompt: str = Field(description="Prompt to interact with the backend.")
-    arg_data: ArgumentData = Field(
-        description="Experiment over graph data.", default=None
-    )
+    arg_data: ArgumentData = Field(description="Experiment over graph data.", default=None)
 
 
 class MultimodalSubgraphExtractionTool(BaseTool):
@@ -59,10 +53,19 @@ class MultimodalSubgraphExtractionTool(BaseTool):
 
     name: str = "subgraph_extraction"
     description: str = "A tool for subgraph extraction based on user's prompt."
-    args_schema: Type[BaseModel] = MultimodalSubgraphExtractionInput
+    args_schema: type[BaseModel] = MultimodalSubgraphExtractionInput
+
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+        # Initialize hardware detection and dynamic library loading
+        object.__setattr__(self, "detector", SystemDetector())
+        object.__setattr__(self, "loader", DynamicLibraryLoader(self.detector))
+        logger.info(
+            "MultimodalSubgraphExtractionTool initialized with %s mode",
+            "GPU" if self.loader.use_gpu else "CPU",
+        )
 
-    def _read_multimodal_files(self,
-                               state: Annotated[dict, InjectedState]) -> df.DataFrame:
+    def _read_multimodal_files(self, state: Annotated[dict, InjectedState]):
         """
         Read the uploaded multimodal files and return a DataFrame.
 
@@ -72,7 +75,7 @@ class MultimodalSubgraphExtractionTool(BaseTool):
         Returns:
             A DataFrame containing the multimodal files.
         """
-        multimodal_df = df.DataFrame({"name": [], "node_type": []})
+        multimodal_df = self.loader.df.DataFrame({"name": [], "node_type": []})
 
         # Loop over the uploaded files and find multimodal files
         logger.log(logging.INFO, "Looping over uploaded files")
@@ -80,8 +83,9 @@ class MultimodalSubgraphExtractionTool(BaseTool):
             # Check if multimodal file is uploaded
             if state["uploaded_files"][i]["file_type"] == "multimodal":
                 # Read the Excel file
-                multimodal_df = pd.read_excel(state["uploaded_files"][i]["file_path"],
-                                                sheet_name=None)
+                multimodal_df = pd.read_excel(
+                    state["uploaded_files"][i]["file_path"], sheet_name=None
+                )
 
         # Check if the multimodal_df is empty
         logger.log(logging.INFO, "Checking if multimodal_df is empty")
@@ -90,20 +94,48 @@ class MultimodalSubgraphExtractionTool(BaseTool):
             logger.log(logging.INFO, "Preparing multimodal_df")
             # Merge all obtained dataframes into a single dataframe
             multimodal_df = pd.concat(multimodal_df).reset_index()
-            multimodal_df = df.DataFrame(multimodal_df)
+            multimodal_df = self.loader.df.DataFrame(multimodal_df)
             multimodal_df.drop(columns=["level_1"], inplace=True)
-            multimodal_df.rename(columns={"level_0": "q_node_type",
-                                        "name": "q_node_name"}, inplace=True)
+            multimodal_df.rename(
+                columns={"level_0": "q_node_type", "name": "q_node_name"}, inplace=True
+            )
             # Since an excel sheet name could not contain a `/`,
             # but the node type can be 'gene/protein' as exists in the PrimeKG
-            multimodal_df["q_node_type"] = multimodal_df["q_node_type"].str.replace('-', '_')
+            multimodal_df["q_node_type"] = multimodal_df["q_node_type"].str.replace("-", "_")
 
         return multimodal_df
 
-    def _prepare_query_modalities(self,
-                                  prompt: dict,
-                                  state: Annotated[dict, InjectedState],
-                                  cfg_db: dict) -> df.DataFrame:
+    def _query_milvus_collection(self, node_type, node_type_df, cfg_db):
+        """Helper method to query Milvus collection for a specific node type."""
+        # Load the collection
+        collection = Collection(
+            name=f"{cfg_db.milvus_db.database_name}_nodes_{node_type.replace('/', '_')}"
+        )
+        collection.load()
+
+        # Query the collection with node names from multimodal_df
+        node_names_series = node_type_df["q_node_name"]
+        q_node_names = getattr(
+            node_names_series, "to_pandas", lambda series=node_names_series: series
+        )().tolist()
+        q_columns = ["node_id", "node_name", "node_type", "feat", "feat_emb", "desc", "desc_emb"]
+        res = collection.query(
+            expr=f"node_name IN [{','.join(f'"{name}"' for name in q_node_names)}]",
+            output_fields=q_columns,
+        )
+        # Convert the embeedings into floats
+        for r_ in res:
+            r_["feat_emb"] = [float(x) for x in r_["feat_emb"]]
+            r_["desc_emb"] = [float(x) for x in r_["desc_emb"]]
+
+        # Convert the result to a DataFrame
+        res_df = self.loader.df.DataFrame(res)[q_columns]
+        res_df["use_description"] = False
+        return res_df
+
+    def _prepare_query_modalities(
+        self, prompt: dict, state: Annotated[dict, InjectedState], cfg_db: dict
+    ):
         """
         Prepare the modality-specific query for subgraph extraction.
 
@@ -118,16 +150,18 @@ class MultimodalSubgraphExtractionTool(BaseTool):
         # Initialize dataframes
         logger.log(logging.INFO, "Initializing dataframes")
         query_df = []
-        prompt_df = df.DataFrame({
-            'node_id': 'user_prompt',
-            'node_name': 'User Prompt',
-            'node_type': 'prompt',
-            'feat': prompt["text"],
-            'feat_emb': prompt["emb"],
-            'desc': prompt["text"],
-            'desc_emb': prompt["emb"],
-            'use_description': True # set to True for user prompt embedding
-        })
+        prompt_df = self.loader.df.DataFrame(
+            {
+                "node_id": "user_prompt",
+                "node_name": "User Prompt",
+                "node_type": "prompt",
+                "feat": prompt["text"],
+                "feat_emb": prompt["emb"],
+                "desc": prompt["text"],
+                "desc_emb": prompt["emb"],
+                "use_description": True,  # set to True for user prompt embedding
+            }
+        )
 
         # Read multimodal files uploaded by the user
         multimodal_df = self._read_multimodal_files(state)
@@ -136,64 +170,44 @@ class MultimodalSubgraphExtractionTool(BaseTool):
         logger.log(logging.INFO, "Prepare query modalities")
         if len(multimodal_df) > 0:
             # Query the Milvus database for each node type in multimodal_df
-            logger.log(logging.INFO, "Querying Milvus database for each node type in multimodal_df")
+            logger.log(
+                logging.INFO,
+                "Querying Milvus database for each node type in multimodal_df",
+            )
             for node_type, node_type_df in multimodal_df.groupby("q_node_type"):
                 print(f"Processing node type: {node_type}")
-
-                # Load the collection
-                collection = Collection(
-                    name=f"{cfg_db.milvus_db.database_name}_nodes_{node_type.replace('/', '_')}"
-                )
-                collection.load()
-
-                # Query the collection with node names from multimodal_df
-                q_node_names =  getattr(node_type_df['q_node_name'],
-                                        "to_pandas",
-                                        lambda: node_type_df['q_node_name'])().tolist()
-                q_columns = ["node_id", "node_name", "node_type",
-                             "feat", "feat_emb", "desc", "desc_emb"]
-                res = collection.query(
-                    expr=f'node_name IN [{','.join(f'"{name}"' for name in q_node_names)}]',
-                    output_fields=q_columns,
-                )
-                # Convert the embeedings into floats
-                for r_ in res:
-                    r_['feat_emb'] = [float(x) for x in r_['feat_emb']]
-                    r_['desc_emb'] = [float(x) for x in r_['desc_emb']]
-
-                # Convert the result to a DataFrame
-                res_df = df.DataFrame(res)[q_columns]
-                res_df["use_description"] = False
-
-                # Append the results to query_df
+                res_df = self._query_milvus_collection(node_type, node_type_df, cfg_db)
                 query_df.append(res_df)
 
             # Concatenate all results into a single DataFrame
             logger.log(logging.INFO, "Concatenating all results into a single DataFrame")
-            query_df = df.concat(query_df, ignore_index=True)
+            query_df = self.loader.df.concat(query_df, ignore_index=True)
 
             # Update the state by adding the the selected node IDs
             logger.log(logging.INFO, "Updating state with selected node IDs")
-            state["selections"] = getattr(query_df,
-                                          "to_pandas",
-                                          lambda: query_df)().groupby(
-                "node_type"
-            )["node_id"].apply(list).to_dict()
+            state["selections"] = (
+                getattr(query_df, "to_pandas", lambda: query_df)()
+                .groupby("node_type")["node_id"]
+                .apply(list)
+                .to_dict()
+            )
 
             # Append a user prompt to the query dataframe
             logger.log(logging.INFO, "Adding user prompt to query dataframe")
-            query_df = df.concat([query_df, prompt_df]).reset_index(drop=True)
+            query_df = self.loader.df.concat([query_df, prompt_df]).reset_index(drop=True)
         else:
             # If no multimodal files are uploaded, use the prompt embeddings
             query_df = prompt_df
 
         return query_df
 
-    def _perform_subgraph_extraction(self,
-                                     state: Annotated[dict, InjectedState],
-                                     cfg: dict,
-                                     cfg_db: dict,
-                                     query_df: pd.DataFrame) -> dict:
+    def _perform_subgraph_extraction(
+        self,
+        state: Annotated[dict, InjectedState],
+        cfg: dict,
+        cfg_db: dict,
+        query_df: pd.DataFrame,
+    ) -> dict:
         """
         Perform multimodal subgraph extraction based on modal-specific embeddings.
 
@@ -208,10 +222,7 @@ class MultimodalSubgraphExtractionTool(BaseTool):
         """
         # Initialize the subgraph dictionary
         subgraphs = []
-        unified_subgraph = {
-            "nodes": [],
-            "edges": []
-        }
+        unified_subgraph = {"nodes": [], "edges": []}
         # subgraphs = {}
         # subgraphs["nodes"] = []
         # subgraphs["edges"] = []
@@ -219,10 +230,22 @@ class MultimodalSubgraphExtractionTool(BaseTool):
         # Loop over query embeddings and modalities
         for q in getattr(query_df, "to_pandas", lambda: query_df)().iterrows():
             logger.log(logging.INFO, "===========================================")
-            logger.log(logging.INFO, "Processing query: %s", q[1]['node_name'])
+            logger.log(logging.INFO, "Processing query: %s", q[1]["node_name"])
             # Prepare the PCSTPruning object and extract the subgraph
             # Parameters were set in the configuration file obtained from Hydra
             # start = datetime.datetime.now()
+            # Get dynamic metric type (overrides any config setting)
+            # Get dynamic metric type (overrides any config setting)
+            has_vector_processing = hasattr(cfg, "vector_processing")
+            if has_vector_processing:
+                dynamic_metrics_enabled = getattr(cfg.vector_processing, "dynamic_metrics", True)
+            else:
+                dynamic_metrics_enabled = False
+            if has_vector_processing and dynamic_metrics_enabled:
+                dynamic_metric_type = self.loader.metric_type
+            else:
+                dynamic_metric_type = getattr(cfg, "search_metric_type", self.loader.metric_type)
+
             subgraph = MultimodalPCSTPruning(
                 topk=state["topk_nodes"],
                 topk_e=state["topk_edges"],
@@ -232,49 +255,51 @@ class MultimodalSubgraphExtractionTool(BaseTool):
                 num_clusters=cfg.num_clusters,
                 pruning=cfg.pruning,
                 verbosity_level=cfg.verbosity_level,
-                use_description=q[1]['use_description'],
-                metric_type=cfg.search_metric_type
-            ).extract_subgraph(q[1]['desc_emb'],
-                               q[1]['feat_emb'],
-                               q[1]['node_type'],
-                               cfg_db)
+                use_description=q[1]["use_description"],
+                metric_type=dynamic_metric_type,  # Use dynamic or config metric type
+                loader=self.loader,  # Pass the loader instance
+            ).extract_subgraph(q[1]["desc_emb"], q[1]["feat_emb"], q[1]["node_type"], cfg_db)
 
             # Append the extracted subgraph to the dictionary
             unified_subgraph["nodes"].append(subgraph["nodes"].tolist())
             unified_subgraph["edges"].append(subgraph["edges"].tolist())
-            subgraphs.append((q[1]['node_name'],
-                              subgraph["nodes"].tolist(),
-                              subgraph["edges"].tolist()))
+            subgraphs.append(
+                (
+                    q[1]["node_name"],
+                    subgraph["nodes"].tolist(),
+                    subgraph["edges"].tolist(),
+                )
+            )
 
             # end = datetime.datetime.now()
             # logger.log(logging.INFO, "Subgraph extraction time: %s seconds",
             #            (end - start).total_seconds())
 
         # Concatenate and get unique node and edge indices
-        unified_subgraph["nodes"] = py.unique(
-            py.concatenate([py.array(list_) for list_ in unified_subgraph["nodes"]])
+        nodes_arrays = [self.loader.py.array(list_) for list_ in unified_subgraph["nodes"]]
+        unified_subgraph["nodes"] = self.loader.py.unique(
+            self.loader.py.concatenate(nodes_arrays)
         ).tolist()
-        unified_subgraph["edges"] = py.unique(
-            py.concatenate([py.array(list_) for list_ in unified_subgraph["edges"]])
+        edges_arrays = [self.loader.py.array(list_) for list_ in unified_subgraph["edges"]]
+        unified_subgraph["edges"] = self.loader.py.unique(
+            self.loader.py.concatenate(edges_arrays)
         ).tolist()
 
-        # Convert the unified subgraph and subgraphs to cudf DataFrames
-        unified_subgraph = df.DataFrame([("Unified Subgraph",
-                                            unified_subgraph["nodes"],
-                                            unified_subgraph["edges"])],
-                                            columns=["name", "nodes", "edges"])
-        subgraphs = df.DataFrame(subgraphs, columns=["name", "nodes", "edges"])
+        # Convert the unified subgraph and subgraphs to DataFrames
+        unified_subgraph = self.loader.df.DataFrame(
+            [("Unified Subgraph", unified_subgraph["nodes"], unified_subgraph["edges"])],
+            columns=["name", "nodes", "edges"],
+        )
+        subgraphs = self.loader.df.DataFrame(subgraphs, columns=["name", "nodes", "edges"])
 
-        # Concate both DataFrames
-        subgraphs = df.concat([unified_subgraph, subgraphs], ignore_index=True)
+        # Concatenate both DataFrames
+        subgraphs = self.loader.df.concat([unified_subgraph, subgraphs], ignore_index=True)
 
         return subgraphs
 
-    def _prepare_final_subgraph(self,
-                                state:Annotated[dict, InjectedState],
-                                subgraph: dict,
-                                cfg: dict,
-                                cfg_db) -> dict:
+    def _prepare_final_subgraph(
+        self, state: Annotated[dict, InjectedState], subgraph: dict, cfg: dict, cfg_db
+    ) -> dict:
         """
         Prepare the subgraph based on the extracted subgraph.
 
@@ -288,94 +313,110 @@ class MultimodalSubgraphExtractionTool(BaseTool):
         Returns:
             A dictionary containing the PyG graph, NetworkX graph, and textualized graph.
         """
-        # Convert the dict to a cudf DataFrame
-        node_colors = {n: cfg.node_colors_dict[k]
-                        for k, v in state["selections"].items() for n in v}
-        color_df = df.DataFrame(list(node_colors.items()), columns=["node_id", "color"])
+        # Convert the dict to a DataFrame
+        node_colors = {
+            n: cfg.node_colors_dict[k] for k, v in state["selections"].items() for n in v
+        }
+        color_df = self.loader.df.DataFrame(list(node_colors.items()), columns=["node_id", "color"])
         # print(color_df)
 
         # Prepare the subgraph dictionary
-        graph_dict = {
-            "name": [],
-            "nodes": [],
-            "edges": [],
-            "text": ""
-        }
+        graph_dict = {"name": [], "nodes": [], "edges": [], "text": ""}
         for sub in getattr(subgraph, "to_pandas", lambda: subgraph)().itertuples(index=False):
-            # Prepare the graph name
-            print(f"Processing subgraph: {sub.name}")
-            print('---')
-            print(sub.nodes)
-            print('---')
-            print(sub.edges)
-            print('---')
-
-            # Prepare graph dataframes
-            # Nodes
-            coll_name = f"{cfg_db.milvus_db.database_name}_nodes"
-            node_coll = Collection(name=coll_name)
-            node_coll.load()
-            graph_nodes = node_coll.query(
-                expr=f'node_index IN [{",".join(f"{n}" for n in sub.nodes)}]',
-                output_fields=['node_id', 'node_name', 'node_type', 'desc']
-            )
-            graph_nodes = df.DataFrame(graph_nodes)
-            graph_nodes.drop(columns=['node_index'], inplace=True)
-            if not color_df.empty:
-                # Merge the color dataframe with the graph nodes
-                graph_nodes = graph_nodes.merge(color_df, on="node_id", how="left")
-            else:
-                graph_nodes["color"] = 'black'  # Default color
-            graph_nodes['color'].fillna('black', inplace=True) # Fill NaN colors with black
-            # Edges
-            coll_name = f"{cfg_db.milvus_db.database_name}_edges"
-            edge_coll = Collection(name=coll_name)
-            edge_coll.load()
-            graph_edges = edge_coll.query(
-                expr=f'triplet_index IN [{",".join(f"{e}" for e in sub.edges)}]',
-                output_fields=['head_id', 'tail_id', 'edge_type']
-            )
-            graph_edges = df.DataFrame(graph_edges)
-            graph_edges.drop(columns=['triplet_index'], inplace=True)
-            graph_edges['edge_type'] = graph_edges['edge_type'].str.split('|')
+            graph_nodes, graph_edges = self._process_subgraph_data(sub, cfg_db, color_df)
 
             # Prepare lists for visualization
             graph_dict["name"].append(sub.name)
-            graph_dict["nodes"].append([(
-                row.node_id,
-                {'hover': "Node Name : " + row.node_name + "\n" +\
-                    "Node Type : " + row.node_type + "\n" +
-                    "Desc : " + row.desc,
-                'click': '$hover',
-                'color': row.color})
-                for row in getattr(graph_nodes,
-                                   "to_pandas",
-                                   lambda: graph_nodes)().itertuples(index=False)])
-            graph_dict["edges"].append([(
-                row.head_id,
-                row.tail_id,
-                {'label': tuple(row.edge_type)})
-                for row in getattr(graph_edges,
-                                   "to_pandas",
-                                   lambda: graph_edges)().itertuples(index=False)])
+            graph_dict["nodes"].append(
+                [
+                    (
+                        row.node_id,
+                        {
+                            "hover": "Node Name : "
+                            + row.node_name
+                            + "\n"
+                            + "Node Type : "
+                            + row.node_type
+                            + "\n"
+                            + "Desc : "
+                            + row.desc,
+                            "click": "$hover",
+                            "color": row.color,
+                        },
+                    )
+                    for row in getattr(
+                        graph_nodes, "to_pandas", lambda graph_nodes=graph_nodes: graph_nodes
+                    )().itertuples(index=False)
+                ]
+            )
+            graph_dict["edges"].append(
+                [
+                    (row.head_id, row.tail_id, {"label": tuple(row.edge_type)})
+                    for row in getattr(
+                        graph_edges, "to_pandas", lambda graph_edges=graph_edges: graph_edges
+                    )().itertuples(index=False)
+                ]
+            )
 
             # Prepare the textualized subgraph
             if sub.name == "Unified Subgraph":
-                graph_nodes = graph_nodes[['node_id', 'desc']]
-                graph_nodes.rename(columns={'desc': 'node_attr'}, inplace=True)
-                graph_edges = graph_edges[['head_id', 'edge_type', 'tail_id']]
-                graph_dict["text"] = (
-                    getattr(graph_nodes, "to_pandas", lambda: graph_nodes)().to_csv(index=False)
-                    + "\n"
-                    + getattr(graph_edges, "to_pandas", lambda: graph_edges)().to_csv(index=False)
-                )
+                graph_nodes = graph_nodes[["node_id", "desc"]]
+                graph_nodes.rename(columns={"desc": "node_attr"}, inplace=True)
+                graph_edges = graph_edges[["head_id", "edge_type", "tail_id"]]
+                nodes_pandas = getattr(
+                    graph_nodes, "to_pandas", lambda graph_nodes=graph_nodes: graph_nodes
+                )()
+                nodes_csv = nodes_pandas.to_csv(index=False)
+                edges_pandas = getattr(
+                    graph_edges, "to_pandas", lambda graph_edges=graph_edges: graph_edges
+                )()
+                edges_csv = edges_pandas.to_csv(index=False)
+                graph_dict["text"] = nodes_csv + "\n" + edges_csv
 
         return graph_dict
 
-    def normalize_vector(self,
-                         v : list) -> list:
+    def _process_subgraph_data(self, sub, cfg_db, color_df):
+        """Helper method to process individual subgraph data."""
+        print(f"Processing subgraph: {sub.name}")
+        print("---")
+        print(sub.nodes)
+        print("---")
+        print(sub.edges)
+        print("---")
+
+        # Prepare graph dataframes - Nodes
+        coll_name = f"{cfg_db.milvus_db.database_name}_nodes"
+        node_coll = Collection(name=coll_name)
+        node_coll.load()
+        graph_nodes = node_coll.query(
+            expr=f"node_index IN [{','.join(f'{n}' for n in sub.nodes)}]",
+            output_fields=["node_id", "node_name", "node_type", "desc"],
+        )
+        graph_nodes = self.loader.df.DataFrame(graph_nodes)
+        graph_nodes.drop(columns=["node_index"], inplace=True)
+        if not color_df.empty:
+            graph_nodes = graph_nodes.merge(color_df, on="node_id", how="left")
+        else:
+            graph_nodes["color"] = "black"
+        graph_nodes["color"] = graph_nodes["color"].fillna("black")
+
+        # Edges
+        coll_name = f"{cfg_db.milvus_db.database_name}_edges"
+        edge_coll = Collection(name=coll_name)
+        edge_coll.load()
+        graph_edges = edge_coll.query(
+            expr=f"triplet_index IN [{','.join(f'{e}' for e in sub.edges)}]",
+            output_fields=["head_id", "tail_id", "edge_type"],
+        )
+        graph_edges = self.loader.df.DataFrame(graph_edges)
+        graph_edges.drop(columns=["triplet_index"], inplace=True)
+        graph_edges["edge_type"] = graph_edges["edge_type"].str.split("|")
+
+        return graph_nodes, graph_edges
+
+    def normalize_vector(self, v: list) -> list:
         """
-        Normalize a vector using CuPy.
+        Normalize a vector using appropriate library (CuPy for GPU, NumPy for CPU).
 
         Args:
             v : Vector to normalize.
@@ -383,9 +424,13 @@ class MultimodalSubgraphExtractionTool(BaseTool):
         Returns:
             Normalized vector.
         """
-        v = py.asarray(v)
-        norm = py.linalg.norm(v)
-        return (v / norm).tolist()
+        if self.loader.normalize_vectors:
+            # GPU mode: normalize the vector
+            v_array = self.loader.py.asarray(v)
+            norm = self.loader.py.linalg.norm(v_array)
+            return (v_array / norm).tolist()
+        # CPU mode: return as-is for COSINE similarity
+        return v
 
     def _run(
         self,
@@ -411,7 +456,8 @@ class MultimodalSubgraphExtractionTool(BaseTool):
         # Load hydra configuration
         with hydra.initialize(version_base=None, config_path="../configs"):
             cfg = hydra.compose(
-                config_name="config", overrides=["tools/multimodal_subgraph_extraction=default"]
+                config_name="config",
+                overrides=["tools/multimodal_subgraph_extraction=default"],
             )
             cfg_db = cfg.app.frontend
             cfg = cfg.tools.multimodal_subgraph_extraction
@@ -431,10 +477,9 @@ class MultimodalSubgraphExtractionTool(BaseTool):
         logger.log(logging.INFO, "_prepare_query_modalities")
         # start = datetime.datetime.now()
         query_df = self._prepare_query_modalities(
-            {"text": prompt,
-             "emb": [self.normalize_vector(
-                 state["embedding_model"].embed_query(prompt)
-                 )]
+            {
+                "text": prompt,
+                "emb": [self.normalize_vector(state["embedding_model"].embed_query(prompt))],
             },
             state,
             cfg_db,
@@ -446,10 +491,7 @@ class MultimodalSubgraphExtractionTool(BaseTool):
         # Perform subgraph extraction
         logger.log(logging.INFO, "_perform_subgraph_extraction")
         # start = datetime.datetime.now()
-        subgraphs = self._perform_subgraph_extraction(state,
-                                                      cfg,
-                                                      cfg_db,
-                                                      query_df)
+        subgraphs = self._perform_subgraph_extraction(state, cfg, cfg_db, query_df)
         # end = datetime.datetime.now()
         # logger.log(logging.INFO, "_perform_subgraph_extraction time: %s seconds",
         #            (end - start).total_seconds())
@@ -458,10 +500,7 @@ class MultimodalSubgraphExtractionTool(BaseTool):
         logger.log(logging.INFO, "_prepare_final_subgraph")
         logger.log(logging.INFO, "Subgraphs extracted: %s", len(subgraphs))
         # start = datetime.datetime.now()
-        final_subgraph = self._prepare_final_subgraph(state,
-                                                      subgraphs,
-                                                      cfg,
-                                                      cfg_db)
+        final_subgraph = self._prepare_final_subgraph(state, subgraphs, cfg, cfg_db)
         # end = datetime.datetime.now()
         # logger.log(logging.INFO, "_prepare_final_subgraph time: %s seconds",
         #            (end - start).total_seconds())
@@ -497,7 +536,8 @@ class MultimodalSubgraphExtractionTool(BaseTool):
 
         # Return the updated state of the tool
         return Command(
-            update=dic_updated_state_for_model | {
+            update=dic_updated_state_for_model
+            | {
                 # update the message history
                 "messages": [
                     ToolMessage(