Update with h2oGPT hash abc352fb7f37932e918f36f428dfb6243219666c

src/gen.py

@@ -2291,19 +2291,22 @@ def evaluate(
         set(langchain_agents).difference(langchain_agents_list)) == 0, "Invalid langchain_agents %s" % langchain_agents
 
     # get db, but also fill db state so return already has my_db_state and dbs filled so faster next query
-    from src.gpt_langchain import get_any_db
-    db = get_any_db(my_db_state, langchain_mode, langchain_mode_paths, langchain_mode_types,
-                    dbs=dbs,
-                    load_db_if_exists=load_db_if_exists,
-                    db_type=db_type,
-                    use_openai_embedding=use_openai_embedding,
-                    hf_embedding_model=hf_embedding_model,
-                    migrate_embedding_model=migrate_embedding_model,
-                    auto_migrate_db=auto_migrate_db,
-                    for_sources_list=True,
-                    verbose=verbose,
-                    n_jobs=n_jobs,
-                    )
+    if langchain_mode != LangChainMode.DISABLED.value:
+        from src.gpt_langchain import get_any_db
+        db = get_any_db(my_db_state, langchain_mode, langchain_mode_paths, langchain_mode_types,
+                        dbs=dbs,
+                        load_db_if_exists=load_db_if_exists,
+                        db_type=db_type,
+                        use_openai_embedding=use_openai_embedding,
+                        hf_embedding_model=hf_embedding_model,
+                        migrate_embedding_model=migrate_embedding_model,
+                        auto_migrate_db=auto_migrate_db,
+                        for_sources_list=True,
+                        verbose=verbose,
+                        n_jobs=n_jobs,
+                        )
+    else:
+        db = None
 
     t_generate = time.time()
     langchain_only_model = base_model in non_hf_types or \