LLMs struggle with Graph RAG, because it's different from vector RAG in
a way that you don't provide the whole context, only the answer and the
LLM has to believe. However, that doesn't really work a lot of the time.
However, if you wrap the context as function response the accuracy is
much better.
btw... `union[LLMChain, Runnable]` is linting fun, that's why so many
ignores
" 'result': 'Anthony Edwards, Meg Ryan, Val Kilmer, Tom Cruise played in Top Gun.'}"
" 'result': 'Tom Cruise, Val Kilmer, Anthony Edwards, and Meg Ryan played in Top Gun.'}"
]
},
"execution_count": 22,
@ -682,13 +682,117 @@
"chain.invoke({\"query\": \"Who played in Top Gun?\"})"
]
},
{
"cell_type": "markdown",
"id": "81093062-eb7f-4d96-b1fd-c36b8f1b9474",
"metadata": {},
"source": [
"## Provide context from database results as tool/function output\n",
"\n",
"You can use the `use_function_response` parameter to pass context from database results to an LLM as a tool/function output. This method improves the response accuracy and relevance of an answer as the LLM follows the provided context more closely.\n",
"_You will need to use an LLM with native function calling support to use this feature_."
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "3fa3f3d5-f7e7-4ca9-8f07-ca22b897f192",
"execution_count": 23,
"id": "2be8f51c-e80a-4a60-ab1c-266450fc17cd",
"metadata": {},
"outputs": [],
"source": []
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"\n",
"\n",
"\u001b[1m> Entering new GraphCypherQAChain chain...\u001b[0m\n",
"chain.invoke({\"query\": \"Who played in Top Gun?\"})"
]
},
{
"cell_type": "markdown",
"id": "48a75785-5bc9-49a7-a41b-88bf3ef9d312",
"metadata": {},
"source": [
"You can provide custom system message when using the function response feature by providing `function_response_system` to instruct the model on how to generate answers.\n",
"\n",
"_Note that `qa_prompt` will have no effect when using `use_function_response`_"
]
},
{
"cell_type": "code",
"execution_count": 24,
"id": "ddf0a61e-f104-4dbb-abbf-e65f3f57dd9a",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"\n",
"\n",
"\u001b[1m> Entering new GraphCypherQAChain chain...\u001b[0m\n",
" 'result': \"Arrr matey! In the film Top Gun, ye be seein' Tom Cruise, Val Kilmer, Anthony Edwards, and Meg Ryan sailin' the high seas of the sky! Aye, they be a fine crew of actors, they be!\"}"
Tomaz Bratanic
4 months ago
committed by
GitHub