Chat bot require grounding

backend/score.py

@@ -481,6 +481,7 @@ async def magic_trek_chat_bot(
     print(messages)
     print(len(messages))
     print(f"document_names = {document_names}")
+    print(f"filter_properties = {filter_properties}")
     logging.info(f"requireGrounding = {requireGrounding}")
     logging.info(f"IAN-TEST called at {datetime.now()}")
     qa_rag_start_time = time.time()

backend/src/QA_integration.py

@@ -397,22 +397,37 @@ def initialize_neo4j_vector(graph, chat_mode_settings):
 
 def create_retriever(neo_db, document_names, chat_mode_settings,search_k, score_threshold,ef_ratio, filter_properties=None):
     if document_names and chat_mode_settings["document_filter"]:
+        # Combine filter_properties with fileName filter
+        base_filter = {'fileName': {'$in': document_names}}
+        if filter_properties:
+            base_filter.update(filter_properties)
+
         retriever = neo_db.as_retriever(
             search_type="similarity_score_threshold",
             search_kwargs={
                 'top_k': search_k,
                 'effective_search_ratio': ef_ratio,
                 'score_threshold': score_threshold,
-                'filter': {'fileName': {'$in': document_names}}
+                'filter': base_filter
             }
         )
         logging.info(f"Successfully created retriever with search_k={search_k}, score_threshold={score_threshold} for documents {document_names}")
     else:
         # note from ian changed the way this is written slightly to match how the upstream team is doing it, passing filter_properties or None to search_kwargs literal.            
+
+        search_kwargs = {
+            'k': search_k, 
+            'score_threshold': score_threshold
+        }
+        if filter_properties is not None:
+            search_kwargs['filter'] = filter_properties
+
         retriever = neo_db.as_retriever(
             search_type="similarity_score_threshold",
-            search_kwargs={'top_k': search_k,'effective_search_ratio': ef_ratio, 'score_threshold': score_threshold, "filter_properties": filter_properties if filter_properties is not None else None }
+            # search_kwargs={'top_k': search_k,'effective_search_ratio': ef_ratio, 'score_threshold': score_threshold, "filter_properties": filter_properties if filter_properties is not None else None }
+            search_kwargs=search_kwargs
         )
+        print(f"search_kwargs = {search_kwargs}")
         logging.info(f"Successfully created retriever with search_k={search_k}, score_threshold={score_threshold}")
     return retriever
 

deploy.sh

@@ -2,10 +2,20 @@
 
 # Variables - Replace with your actual values
 
-AWS_REGION="us-east-1"
-AWS_ENV="dev"
-AWS_ACCOUNT_ID="288761770129" # 288 is graphbuilder-dev
+
+# uncomment htis for 'dev' (labs llm-graphbuilder)
+# AWS_REGION="us-east-1"
+# AWS_ENV="dev"
+# AWS_ACCOUNT_ID="288761770129" # 288 is graphbuilder-dev
+
+# uncomment this for 'metrix-demo-dev' (metrix-demo-dev)
+AWS_REGION="us-east-2"
+AWS_ENV="metrix-demo-dev"
+AWS_ACCOUNT_ID="860839672899"
+
 ECR_REPOSITORY="graphbuilder-docker-repo-$AWS_ENV"
+ENV_FILE="./${AWS_ENV}.env"
+echo "Using ENV_FILE $ENV_FILE"
 TAG="latest"  # You can change this if you need a specific tag
 
 folder_name=$(basename "$PWD")
@@ -14,9 +24,16 @@ echo "Using AWS_PROFILE $AWS_PROFILE"
 echo "Using AWS_ENV $AWS_ENV"
 echo "Using ECR_REPOSITORY $ECR_REPOSITORY"
 
-# 1. Build Docker images using Docker Compose
+# 1. Load environment variables from .env file
+echo "Loading environment variables from $ENV_FILE..."
+set -a  # automatically export all variables
+source $ENV_FILE
+set +a  # turn off automatic export
+
+# 2. Build Docker images using Docker Compose
 echo "Building Docker images..."
-docker-compose build --no-cache --progress=plain
+# docker-compose build --no-cache --progress=plain
+docker-compose build
 
 # 2. Authenticate Docker to AWS ECR
 echo "Authenticating to AWS ECR..."