mirror of
https://github.com/run-llama/llama_cloud_services.git
synced 2026-07-01 21:44:37 -04:00
Logan
1112 lines
53 KiB
Plaintext
1112 lines
53 KiB
Plaintext
{
|
||
"cells": [
|
||
{
|
||
"cell_type": "markdown",
|
||
"id": "6a728deb-28da-4064-8e66-bef416022207",
|
||
"metadata": {},
|
||
"source": [
|
||
"# Knowledge Graph Agent with LlamaParse\n",
|
||
"\n",
|
||
"<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/parse/knowledge_graphs/kg_agent.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
|
||
"\n",
|
||
"Here we build a knowledge graph agent over the SF 2023 Budget Proposal. We use LlamaIndex abstractions to construct a knowledge graph, and we store the property graph in neo4j. We then build an agent that can interact with the knowledge graph as a tool.\n",
|
||
"\n",
|
||
"Status:\n",
|
||
"| Last Executed | Version | State |\n",
|
||
"|---------------|---------|------------|\n",
|
||
"| Before Feb 2025 | N/A | Deprecated |"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"id": "0facb0b9",
|
||
"metadata": {},
|
||
"source": [
|
||
"> **⚠️ DEPRECATION NOTICE**>> This example uses the deprecated `llama-cloud-services` package, which will be maintained until **May 1, 2026**.>> **Please migrate to:**> - **Python**: `pip install llama-cloud>=1.0` ([GitHub](https://github.com/run-llama/llama-cloud-py))> - **New Package Documentation**: https://docs.cloud.llamaindex.ai/>> The new package provides the same functionality with improved performance and support."
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"id": "e8db8ac2-5221-44de-a53e-cb5ab37ac8f5",
|
||
"metadata": {},
|
||
"source": [
|
||
"## Setup (Installs, Data, Models)"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "989d1cb5-5464-4d9d-ac1e-6e16276d3698",
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"!pip install llama-index\n",
|
||
"!pip install llama-index-core==0.10.42\n",
|
||
"!pip install llama-index-embeddings-openai\n",
|
||
"!pip install llama-index-postprocessor-flag-embedding-reranker\n",
|
||
"!pip install git+https://github.com/FlagOpen/FlagEmbedding.git\n",
|
||
"!pip install llama-index-graph-stores-neo4j\n",
|
||
"!pip install llama-cloud-services"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "580ecfc2-b082-4c4e-910a-7f88e8137aad",
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"import nest_asyncio\n",
|
||
"\n",
|
||
"nest_asyncio.apply()"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "c86e95e2-0fdf-4bb1-bf2e-b33af13ac7ef",
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"import os\n",
|
||
"\n",
|
||
"# API access to llama-cloud\n",
|
||
"# os.environ[\"LLAMA_CLOUD_API_KEY\"] = \"llx-\""
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"id": "d6f683f2-a41e-4975-843c-435407132f0e",
|
||
"metadata": {},
|
||
"source": [
|
||
"#### Setup Model\n",
|
||
"\n",
|
||
"Here we use gpt-4o and default OpenAI embeddings."
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "d91854ee-d57a-4d7b-bcc8-c6bd7214fe84",
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"from llama_index.llms.openai import OpenAI\n",
|
||
"from llama_index.embeddings.openai import OpenAIEmbedding\n",
|
||
"from llama_index.core import Settings\n",
|
||
"\n",
|
||
"llm = OpenAI(model=\"gpt-4o\")\n",
|
||
"embed_model = OpenAIEmbedding(model=\"text-embedding-3-small\")\n",
|
||
"\n",
|
||
"Settings.llm = llm\n",
|
||
"Settings.embed_model = embed_model"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"id": "5bcf33f7-b195-444d-9355-ab47c91be6ad",
|
||
"metadata": {},
|
||
"source": [
|
||
"#### Load Data\n",
|
||
"\n",
|
||
"Here we load the 2023 Budget PDF and parse it with LlamaParse."
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "24e7d3f1-012d-4964-8130-e9a386b70996",
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"!mkdir data\n",
|
||
"!wget \"https://www.dropbox.com/scl/fi/vip161t63s56vd94neqlt/2023-CSF_Proposed_Budget_Book_June_2023_Master_Web.pdf?rlkey=hemoce3w1jsuf6s2bz87g549i&dl=0\" -O data/budget_2023.pdf"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "96d13f0b-1749-4c06-a4f7-b6f885db04d3",
|
||
"metadata": {},
|
||
"outputs": [
|
||
{
|
||
"name": "stdout",
|
||
"output_type": "stream",
|
||
"text": [
|
||
"Started parsing the file under job_id a7bc360f-1625-4fb7-a950-7531a8b3447e\n"
|
||
]
|
||
}
|
||
],
|
||
"source": [
|
||
"from llama_cloud_services import LlamaParse\n",
|
||
"\n",
|
||
"docs = LlamaParse(result_type=\"text\").load_data(\"./data/budget_2023.pdf\")"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "9afcf38e-3c7e-4b48-ae23-61b33f1a448a",
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"from copy import deepcopy\n",
|
||
"from llama_index.core.schema import TextNode, Document\n",
|
||
"from llama_index.core import VectorStoreIndex\n",
|
||
"\n",
|
||
"\n",
|
||
"def get_sub_docs(docs):\n",
|
||
" \"\"\"Split docs into pages, by separator.\"\"\"\n",
|
||
" sub_docs = []\n",
|
||
" for doc in docs:\n",
|
||
" doc_chunks = doc.text.split(\"\\n---\\n\")\n",
|
||
" for doc_chunk in doc_chunks:\n",
|
||
" sub_doc = Document(\n",
|
||
" text=doc_chunk,\n",
|
||
" metadata=deepcopy(doc.metadata),\n",
|
||
" )\n",
|
||
" sub_docs.append(sub_doc)\n",
|
||
"\n",
|
||
" return sub_docs"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "d672fcaa-9e49-40a2-9a0b-e709351e840f",
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"# this will split into pages\n",
|
||
"sub_docs = get_sub_docs(docs)"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"id": "e55bcae5-43ba-4363-9109-d8080d19ce5a",
|
||
"metadata": {},
|
||
"source": [
|
||
"#### Initialize Graph Store\n",
|
||
"\n",
|
||
"Here we use Neo4j but you can also use our other integrations like Nebula (see an [example notebook](https://github.com/run-llama/llama_index/blob/main/docs/docs/examples/property_graph/property_graph_advanced.ipynb))."
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"id": "b86a0e7b-bc15-45ed-b1f7-b4c82d16b815",
|
||
"metadata": {},
|
||
"source": [
|
||
"To launch Neo4j locally, first ensure you have docker installed. Then, you can launch the database with the following docker command\n",
|
||
"\n",
|
||
"```bash\n",
|
||
"docker run \\\n",
|
||
" -p 7474:7474 -p 7687:7687 \\\n",
|
||
" -v $PWD/data:/data -v $PWD/plugins:/plugins \\\n",
|
||
" --name neo4j-apoc \\\n",
|
||
" -e NEO4J_apoc_export_file_enabled=true \\\n",
|
||
" -e NEO4J_apoc_import_file_enabled=true \\\n",
|
||
" -e NEO4J_apoc_import_file_use__neo4j__config=true \\\n",
|
||
" -e NEO4JLABS_PLUGINS=\\[\\\"apoc\\\"\\] \\\n",
|
||
" neo4j:latest\n",
|
||
"```\n",
|
||
"\n",
|
||
"From here, you can open the db at [http://localhost:7474/](http://localhost:7474/). On this page, you will be asked to sign in. Use the default username/password of `neo4j` and `neo4j`.\n",
|
||
"\n",
|
||
"Once you login for the first time, you will be asked to change the password.\n",
|
||
"\n",
|
||
"After this, you are ready to create your first property graph!"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "e25ed865-b78e-4856-9473-7123d9924d46",
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"from llama_index.graph_stores.neo4j import Neo4jPGStore\n",
|
||
"\n",
|
||
"graph_store = Neo4jPGStore(\n",
|
||
" username=\"neo4j\",\n",
|
||
" password=\"llamaindex\",\n",
|
||
" url=\"bolt://localhost:7687\",\n",
|
||
")\n",
|
||
"vec_store = None"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"id": "10723825-328f-4175-ad85-637b0c28262c",
|
||
"metadata": {},
|
||
"source": [
|
||
"## Construct Knowledge Graph, Get Retrievers\n",
|
||
"\n",
|
||
"This section shows you how to construct the knowledge graph over the existing documents.\n",
|
||
"\n",
|
||
"**Note**: we have the default extractors (implicit path, simple llm path) configured. You can also choose to use a pre-defined schema as mentioned in this [notebook](https://github.com/run-llama/llama_index/blob/main/docs/docs/examples/property_graph/property_graph_advanced.ipynb)."
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "6b3d81c4-95a8-409b-a448-ac18e193effc",
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"from llama_index.core.indices.property_graph import (\n",
|
||
" ImplicitPathExtractor,\n",
|
||
" SimpleLLMPathExtractor,\n",
|
||
")\n",
|
||
"from llama_index.core import PropertyGraphIndex\n",
|
||
"from llama_index.llms.openai import OpenAI\n",
|
||
"from llama_index.embeddings.openai import OpenAIEmbedding"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "d4debe49-e2a1-4092-b0e7-cc5e6604fbef",
|
||
"metadata": {},
|
||
"outputs": [
|
||
{
|
||
"name": "stderr",
|
||
"output_type": "stream",
|
||
"text": [
|
||
"Parsing nodes: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████| 362/362 [00:00<00:00, 1051.95it/s]\n",
|
||
"Extracting implicit paths: 100%|███████████████████████████████████████████████████████████████████████████████████████| 438/438 [00:00<00:00, 99701.79it/s]\n",
|
||
"Extracting paths from text: 100%|█████████████████████████████████████████████████████████████████████████████████████████| 438/438 [03:53<00:00, 1.87it/s]\n",
|
||
"Generating embeddings: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████| 5/5 [00:01<00:00, 2.89it/s]\n",
|
||
"Generating embeddings: 100%|████████████████████████████████████████████████████████████████████████████████████████████████| 62/62 [00:01<00:00, 38.10it/s]\n"
|
||
]
|
||
}
|
||
],
|
||
"source": [
|
||
"index = PropertyGraphIndex.from_documents(\n",
|
||
" sub_docs,\n",
|
||
" embed_model=OpenAIEmbedding(model_name=\"text-embedding-3-small\"),\n",
|
||
" kg_extractors=[\n",
|
||
" ImplicitPathExtractor(),\n",
|
||
" SimpleLLMPathExtractor(\n",
|
||
" llm=OpenAI(model=\"gpt-3.5-turbo\", temperature=0.3),\n",
|
||
" num_workers=4,\n",
|
||
" max_paths_per_chunk=10,\n",
|
||
" ),\n",
|
||
" ],\n",
|
||
" property_graph_store=graph_store,\n",
|
||
" show_progress=True,\n",
|
||
")"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "ffc29e45-89c1-48cf-b221-c4e117cda630",
|
||
"metadata": {},
|
||
"outputs": [
|
||
{
|
||
"name": "stderr",
|
||
"output_type": "stream",
|
||
"text": [
|
||
"Extracting implicit paths: 0it [00:00, ?it/s]\n",
|
||
"Extracting paths from text: 0it [00:00, ?it/s]\n",
|
||
"Generating embeddings: 0it [00:00, ?it/s]\n",
|
||
"Generating embeddings: 0it [00:00, ?it/s]\n"
|
||
]
|
||
}
|
||
],
|
||
"source": [
|
||
"# run this if index is already loaded\n",
|
||
"index = PropertyGraphIndex.from_existing(\n",
|
||
" graph_store,\n",
|
||
" embed_model=OpenAIEmbedding(model_name=\"text-embedding-3-small\"),\n",
|
||
" kg_extractors=[\n",
|
||
" ImplicitPathExtractor(),\n",
|
||
" SimpleLLMPathExtractor(\n",
|
||
" llm=OpenAI(model=\"gpt-3.5-turbo\", temperature=0.3),\n",
|
||
" num_workers=4,\n",
|
||
" max_paths_per_chunk=10,\n",
|
||
" ),\n",
|
||
" ],\n",
|
||
" show_progress=True,\n",
|
||
")"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"id": "20cf723a-5353-4f97-95b2-c9cde940c583",
|
||
"metadata": {},
|
||
"source": [
|
||
"The constructed knowledge graph should look something like this\n",
|
||
""
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"id": "0c1f0f65-2d15-471c-9e74-6cebcecee1f4",
|
||
"metadata": {},
|
||
"source": [
|
||
"#### Define Vector Retriever\n",
|
||
"\n",
|
||
"Here we define our vector context retriever - it returns initial nodes via vector search, and traverses the relations to pull in more nodes/context."
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "3d4d4ce1-64df-41cd-9f36-49c779fdc2b3",
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"from llama_index.core.indices.property_graph import VectorContextRetriever\n",
|
||
"\n",
|
||
"kg_retriever = VectorContextRetriever(\n",
|
||
" index.property_graph_store,\n",
|
||
" embed_model=OpenAIEmbedding(model_name=\"text-embedding-3-small\"),\n",
|
||
" similarity_top_k=2,\n",
|
||
" path_depth=1,\n",
|
||
" # include_text=False,\n",
|
||
" include_text=True,\n",
|
||
")"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "ca3cd98b-1556-4ae7-8a24-519adb48beb6",
|
||
"metadata": {},
|
||
"outputs": [
|
||
{
|
||
"name": "stdout",
|
||
"output_type": "stream",
|
||
"text": [
|
||
"3\n",
|
||
">> IDX: 0, Here are some facts extracted from the provided text:\n",
|
||
"\n",
|
||
"Mayor's budget -> Includes -> Key changes\n",
|
||
"\n",
|
||
"first responders to petition for an individual to enter\n",
|
||
"the programs. In these procedures, a CARE Plan is\n",
|
||
"established, and a judge can use court orders with\n",
|
||
"support such as short-term stabilization medications\n",
|
||
"and beds, as well as wellness and recovery offerings.\n",
|
||
"The Mayor’s proposed budget includes funding\n",
|
||
"for engagement and assessment staff, new City\n",
|
||
"attorneys dedicated to CARE Court implementation,\n",
|
||
"increased capacity for treatment and housing, and\n",
|
||
"outreach and educational efforts.\n",
|
||
"Improvements at Laguna Honda Hospital\n",
|
||
"Beyond behavioral health, this budget makes\n",
|
||
"investments in DPH’s budget for Laguna Honda\n",
|
||
"Hospital, which is actively working towards gaining\n",
|
||
"recertification with the Centers for Medicare\n",
|
||
"and Medicaid Services (CMS). DPH is currently\n",
|
||
"implementing the action plan submitted to CMS,\n",
|
||
"and it represents a significant facility-wide effort\n",
|
||
"and includes hundreds of process improvements.\n",
|
||
"The Mayor’s proposed budget includes over $3.5\n",
|
||
"million of new annual investment to support the\n",
|
||
"implementation of the action plan and sustain\n",
|
||
"the improvements, including staffing in key areas,\n",
|
||
"including education and training, patient care\n",
|
||
"experience, medication management, and leadership\n",
|
||
"within the San Francisco Health Network.\n",
|
||
"\n",
|
||
"Economic Recovery\n",
|
||
"The Mayor’s proposed FY 2023-24 and FY 2024-\n",
|
||
"25 budget invests $24.4 million over the two years\n",
|
||
"in support of the Roadmap for Downtown San\n",
|
||
"Francisco’s Future, and broadly supports economic\n",
|
||
"recovery across the entire City. While critical\n",
|
||
"components around recovery include investments\n",
|
||
"in public safety and street conditions, there are also\n",
|
||
"targeted improvements and programs to support a\n",
|
||
"thriving economy, both downtown and throughout\n",
|
||
"the City’s neighborhood commercial corridors.\n",
|
||
"Providing Tax Relief and Incentives\n",
|
||
"To keep existing businesses stable and to recruit new\n",
|
||
"businesses, the Mayor’s budget includes key changes\n",
|
||
"to the City’s business taxes. In November 2020, San\n",
|
||
"20\n",
|
||
" 20 EXECUTIVE SUMMARY\n",
|
||
" EXECUTIVE SUMMARY\n",
|
||
"\n",
|
||
"\n",
|
||
"Francisco voters passed Proposition F, which phased\n",
|
||
"out the payroll expense tax, while gradually increasing\n",
|
||
"gross receipts tax for businesses across most\n",
|
||
"industries. The proposition also delayed gross receipts\n",
|
||
"tax increases until 2023 and 2024 for industries hit\n",
|
||
"hardest by the pandemic. In March 2023, the Mayor\n",
|
||
"introduced legislation to further delay tax increases\n",
|
||
"for maintenance and laundry businesses, retail trade,\n",
|
||
"food services, manufacturing, accommodations, arts,\n",
|
||
"entertainment, and recreation until 2025 and 2026.\n",
|
||
"The proposed budget includes revenue assumptions\n",
|
||
"aligned with these tax changes.\n",
|
||
"The City must also attract new businesses to fill its\n",
|
||
"office vacancies, support customer-serving businesses,\n",
|
||
"and bolster future revenue through gross receipts\n",
|
||
"tax, property tax, and contributions to other revenue\n",
|
||
"sources. The Mayor’s legislation will offer a discount\n",
|
||
"for up to three years on the office-based gross\n",
|
||
"receipts tax for new offices locating in San Francisco\n",
|
||
"in the information, administrative and support\n",
|
||
"services, financial services, insurance, professional\n",
|
||
"scientific and technical services industries.\n",
|
||
"\n",
|
||
"Finally, the budget proposes a change to the\n",
|
||
"Commercial Rent Tax, which was passed by voters in\n",
|
||
"June 2018. The change seeks to pause the collection\n",
|
||
"revenues on sub-leases of commercial spaces through\n",
|
||
"2029 to ensure commercial properties are only\n",
|
||
"subject to a single commercial rent tax, rather than\n",
|
||
"also being taxed for sub-leasing the space. Profits on\n",
|
||
"subleases will continue to be collected.\n",
|
||
"Supporting Small Businesses\n",
|
||
"The Mayor’s proposed budget continues the small\n",
|
||
"business grant program, providing $5 million in direct\n",
|
||
"grants to help small businesses across the city to\n",
|
||
"stabilize, scale, and adapt business models to changed\n",
|
||
"conditions. The program will target businesses in\n",
|
||
"commercial corridors that have experienced the\n",
|
||
"highest drop in sales tax to fill vacancies, or expand\n",
|
||
"into new storefronts, while providing business\n",
|
||
"assistance to improve operations and renegotiate\n",
|
||
"leases\n",
|
||
">> IDX: 1, Here are some facts extracted from the provided text:\n",
|
||
"\n",
|
||
"Mayor's proposed budget -> Expands investments within -> San francisco police department\n",
|
||
"Mayor's proposed budget -> Includes -> Funding\n",
|
||
"Mayor's proposed budget -> Invests in -> Facilities maintenance\n",
|
||
"Mayor's proposed budget -> Includes -> $32.0 million\n",
|
||
"Mayor's proposed budget -> Includes -> Mental health services act funds\n",
|
||
"Mayor's proposed budget -> Invests -> $24.4 million\n",
|
||
"Mayor's proposed budget -> Invests in -> Stationary engineers\n",
|
||
"Mayor's proposed budget -> Includes -> $17.7 million\n",
|
||
"Mayor's proposed budget -> Leverages -> Ocoh\n",
|
||
"Mayor's proposed budget -> Makes investments to -> Two positions\n",
|
||
"Mayor's proposed budget -> Makes investments in -> Priority areas\n",
|
||
"Mayor's proposed budget -> Continues -> Work\n",
|
||
"Mayor's proposed budget -> Includes -> $1.2 million of grant funding\n",
|
||
"Mayor's proposed budget -> Maintain -> Programs\n",
|
||
"Mayor's proposed budget -> Includes -> $0.7 million\n",
|
||
"Mayor's proposed budget -> Includes -> New funding to support efforts\n",
|
||
"Mayor's proposed budget -> Includes -> Resources\n",
|
||
"Mayor's proposed budget -> Makes investments to -> Continue supporting\n",
|
||
"Mayor's proposed budget -> Includes -> Anticipated grant\n",
|
||
"Mayor's proposed budget -> Invest in -> Expansion of recovery programs\n",
|
||
"Mayor's proposed budget -> Includes -> $0.6 million\n",
|
||
"Mayor's proposed budget -> Invests in -> Building projects\n",
|
||
"\n",
|
||
"CalAIM Expansion for People At-Risk of\n",
|
||
"Institutionalization and Justice-Involved\n",
|
||
"People\n",
|
||
"The State of California is continuing its multi-year\n",
|
||
"roll out of California Advancing and Innovating\n",
|
||
"Medi-Cal (CalAIM), a new framework that\n",
|
||
"encompasses a broad-based delivery system,\n",
|
||
"program, and payment reform across the Medi-Cal\n",
|
||
"program with a whole-person care approach. For\n",
|
||
"the proposed FY 2023-24 and FY 2024-25 budget,\n",
|
||
"the focus is on the roll-out of expanded benefits\n",
|
||
"to people at risk of long-term institutionalization\n",
|
||
"and justice-system involved people exiting jail.\n",
|
||
"As CalAIM focuses on stabilizing patients in\n",
|
||
"community settings as much as possible, the\n",
|
||
"enhanced care management (ECM) benefit allows\n",
|
||
"for Medi-Cal to pay for hands-on support to\n",
|
||
"address both the clinical and non-clinical needs of\n",
|
||
"medically complex patients to keep them out of\n",
|
||
"institutions.\n",
|
||
"\n",
|
||
"Health Equity Investments through the\n",
|
||
"Mental Health Services Act\n",
|
||
"The Mayor’s proposed budget includes $32.0\n",
|
||
"million in FY 2022-23 and $17.7 million of ongoing\n",
|
||
"\n",
|
||
"\n",
|
||
"in additional Mental Health Services Act (MHSA)\n",
|
||
"funds. These funds will be used to ensure the\n",
|
||
"continuity of existing MHSA programming; support\n",
|
||
"new, innovative and culturally congruent services\n",
|
||
"to meet the pressing needs of the Black/African\n",
|
||
"American community and sustain a pilot effort to\n",
|
||
"provide mental health support for Black mothers.\n",
|
||
"New initiatives in this budget include $15.0 million\n",
|
||
"for a three-year pilot from FY 2023-24 through FY\n",
|
||
"2025-26 with the Dream Keeper Initiative to create\n",
|
||
"a talk therapy, telehealth program for people in San\n",
|
||
"Francisco, with a particular focus on Black/African\n",
|
||
"American residents.\n",
|
||
"Fee for Service Transition\n",
|
||
"\n",
|
||
"In January 2023, the San Francisco Health PlanNumber of Calls(SFHP) and Zuckerberg San Francisco General\n",
|
||
"\n",
|
||
"Hospital (ZSFG) expanded its use of the fee-for-\n",
|
||
"service model to maximize revenues. By expanding\n",
|
||
"the use of fee-for-service, ZSFG will recover\n",
|
||
"more funding from the SFHP and the State, while\n",
|
||
"maintaining quality care for its patients. DPH\n",
|
||
"projects $36.7 M in additional revenues in FY 2023-\n",
|
||
"24 and $36.9 M in FY 2024-25 as result of the shift\n",
|
||
"that is included in the Mayor’s proposed budget.\n",
|
||
"\n",
|
||
"\n",
|
||
"\n",
|
||
"\n",
|
||
" 3,000\n",
|
||
"\n",
|
||
"\n",
|
||
" STREET RESPONSE TEAM. The goals 2,500\n",
|
||
" of the San Francisco Street Overdose\n",
|
||
" Response Team (SORT) are to reduce the 2,000\n",
|
||
" risk of opioid-related death of individuals\n",
|
||
" who have recently experienced an 1,500\n",
|
||
" overdose, contribute to an overall reduction\n",
|
||
" in overdose deaths through referrals and\n",
|
||
" care coordination with community-based 1,000\n",
|
||
" organziations, and to provide support to\n",
|
||
" people who have survived any overdose. 500\n",
|
||
"\n",
|
||
"\n",
|
||
" 0 Calls Handled Calls Including Calls That Include Clients Who\n",
|
||
" by SORT an Overdose Buprennorphrine StartsAccepted Harm\n",
|
||
" Reduction Supplies\n",
|
||
" Types of Calls\n",
|
||
"\n",
|
||
"\n",
|
||
" PUBLIC HEALTH 267\n",
|
||
">> IDX: 2, Here are some facts extracted from the provided text:\n",
|
||
"\n",
|
||
"Input -> Carefully considered in formulating -> Mayor's proposed budget\n",
|
||
"\n",
|
||
"Key Participants\n",
|
||
"• Residents provide direction for and commentary\n",
|
||
" on budget priorities throughout the annual budget\n",
|
||
" process. Input from residents through virtual\n",
|
||
" feedback forms, stakeholder working groups\n",
|
||
" convened by the Mayor’s Office, public budget\n",
|
||
" hearings, and communication with elected officials\n",
|
||
" are all carefully considered in formulating the\n",
|
||
" Mayor’s proposed budget.\n",
|
||
"• City departments prioritize needs and present\n",
|
||
" balanced budgets for review and analysis by the\n",
|
||
" Mayor’s Office of Public Policy and Finance.\n",
|
||
"• The multi-year budget projections described in the\n",
|
||
" previous section as well as the Capital Planning\n",
|
||
" Committee (CPC) and Committee on Information\n",
|
||
" Technology (COIT) provide guidance to the Mayor’s\n",
|
||
" Office on both long-term fiscal trends as well as\n",
|
||
" citywide priorities for capital and IT investments.\n",
|
||
"• The Mayor, with the assistance of the Mayor’s\n",
|
||
" Office of Public Policy and Finance, prepares\n",
|
||
" and submits a balanced budget to the Board of\n",
|
||
" Supervisors on an annual basis.\n",
|
||
"• The Board of Supervisors is the City’s legislative\n",
|
||
" body and is responsible for amending and\n",
|
||
" approving the Mayor’s proposed budget. The\n",
|
||
" Board’s Budget and Legislative Analyst also\n",
|
||
" participates in reviews of city spending and\n",
|
||
" financial projections and makes recommendations\n",
|
||
" to the Board on budget modifications.\n",
|
||
"• The Controller is the City’s Chief Financial Officer\n",
|
||
" and is responsible for projecting available revenue\n",
|
||
" to fund city operations and investments in both\n",
|
||
" the near- and long-term. In addition, the City\n",
|
||
" Services Auditor Division of the Controller’s Office\n",
|
||
" is responsible for working with departments to\n",
|
||
" develop, improve, and evaluate their performance\n",
|
||
" standards.\n",
|
||
"\n",
|
||
"Calendar and Process\n",
|
||
"Beginning in September and concluding in July, the\n",
|
||
"annual budget cycle can be divided into three major\n",
|
||
"stages (see calendar at the end of this section):\n",
|
||
"\n",
|
||
"50 BUDGET PROCESS\n",
|
||
"\n",
|
||
"\n",
|
||
"\n",
|
||
"• Budget Preparation: budget development and\n",
|
||
" submission to the Board of Supervisors.\n",
|
||
"• Approval: budget review and enactment by the\n",
|
||
" Board of Supervisors and budget signing by the\n",
|
||
" Mayor\n",
|
||
"• Implementation: department execution and budget\n",
|
||
" adjustments.\n",
|
||
"\n",
|
||
"Budget Preparation\n",
|
||
"Preliminary projections of Enterprise and General Fund\n",
|
||
"revenues for the next fiscal year by the Controller’s\n",
|
||
"Office and Mayor’s Office staff begin in September.\n",
|
||
"Around this time, many departments begin budget\n",
|
||
"planning to allow adequate input from oversight\n",
|
||
"commissions and the public. In December, budget\n",
|
||
"instructions are issued by the Mayor’s Office and\n",
|
||
"the Controller’s Office with detailed guidance on the\n",
|
||
"preparation of department budget requests. The\n",
|
||
"instructions contain a financial outlook, policy goals,\n",
|
||
"and guidelines as well as technical instructions.\n",
|
||
"Three categories of budgets are prepared:\n",
|
||
"• General Fund department budgets: General Fund\n",
|
||
" departments rely in whole or in part on discretionary\n",
|
||
" revenue comprised primarily of local taxes such as\n",
|
||
" property, sales, payroll, and other taxes. The Mayor\n",
|
||
" introduces the proposed General Fund budget to\n",
|
||
" the Board of Supervisors on June 1.\n",
|
||
"\n",
|
||
"• Enterprise department budgets: Enterprise\n",
|
||
" departments generate non-discretionary revenue\n",
|
||
" primarily from charges for services that are used\n",
|
||
" to support operations. The Mayor introduces\n",
|
||
" the proposed Enterprise budgets to the Board of\n",
|
||
" Supervisors on May 1.\n",
|
||
"• Capital and IT budgets: Capital and IT budget\n",
|
||
" requests are submitted to the CPC and COIT\n",
|
||
" for review. The recommendations for each\n",
|
||
" committee are taken into account during the budget\n",
|
||
" preparation process. The City’s Ten-Year Capital\n",
|
||
" Plan is brought before the Board of Supervisors and\n",
|
||
" Mayor for approval concurrently with the General\n",
|
||
" Fund and Enterprise department budgets.\n"
|
||
]
|
||
}
|
||
],
|
||
"source": [
|
||
"nodes = kg_retriever.retrieve(\n",
|
||
" \"Give me all the programs that the mayor's budget includes\"\n",
|
||
")\n",
|
||
"# nodes = kg_retriever.retrieve('san francisco')\n",
|
||
"print(len(nodes))\n",
|
||
"for idx, node in enumerate(nodes):\n",
|
||
" print(f\">> IDX: {idx}, {node.get_content()}\")"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"id": "1e0867c1-dd3d-4e84-93a3-30b4ec861024",
|
||
"metadata": {},
|
||
"source": [
|
||
"## Build Baseline Vector Index\n",
|
||
"\n",
|
||
"We also build a \"baseline\" vector index. This follows the \"naive\" RAG pipeline approach of chunking and vector embedding. We use this as a comparison point."
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "18b6e528-e1af-41a0-9f04-909e24f1584b",
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"from llama_index.core import VectorStoreIndex\n",
|
||
"from llama_index.core.query_engine import RetrieverQueryEngine\n",
|
||
"\n",
|
||
"base_index = VectorStoreIndex.from_documents(sub_docs, embed_model=embed_model)\n",
|
||
"base_retriever = base_index.as_retriever(similarity_top_k=2)\n",
|
||
"base_query_engine = RetrieverQueryEngine(base_retriever)"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "573bef31-0194-4532-b0b1-ea3e84e3890c",
|
||
"metadata": {},
|
||
"outputs": [
|
||
{
|
||
"name": "stdout",
|
||
"output_type": "stream",
|
||
"text": [
|
||
"The mayor's budget includes the following programs:\n",
|
||
"\n",
|
||
"1. Ongoing programmatic support for all districts.\n",
|
||
"2. One-time $5.0 million and $250,000 ongoing investment in priority community-based organization needs, including capital and infrastructure, and public safety.\n",
|
||
"3. Support for the Mayor’s Office Administration, which advances Mayoral priorities through policy and budget development, communications, and advocacy.\n",
|
||
"4. Financial Capability Services.\n",
|
||
"5. Nonprofit Capacity Building.\n",
|
||
"6. Eviction Prevention and Housing Stabilization Services.\n",
|
||
"7. Community and Housing Place-Based Services.\n",
|
||
"8. Civil Legal Services.\n",
|
||
"9. Supportive Housing for Persons with HIV/AIDS.\n",
|
||
"10. Community, Coalition, and Cultural District Building.\n",
|
||
"11. Rental and Homeownership Counseling.\n",
|
||
"12. Capital Projects.\n",
|
||
"13. Housing Development Grants.\n",
|
||
"14. Creation of permanently affordable housing.\n",
|
||
"15. Foster healthy communities and neighborhoods.\n",
|
||
"16. Improve access to affordable housing.\n",
|
||
"17. Preserve affordable housing.\n",
|
||
"18. Promote self-sufficiency and protect rights.\n",
|
||
"19. Investments in capital and information technology, including urgent repairs and crucial projects across the park system and ADA needs.\n",
|
||
"20. Replacement of critical City systems like the Computer Aided Dispatch system and the Property Tax System.\n",
|
||
"21. Funding for the Sheriff’s Jail Management System, JUSTIS Data Center of Excellence, Infrastructure Modernization, and digital accessibility.\n",
|
||
"22. New projects like the replacement of the legacy Legislative Management System and a new platform for the Empty Homes Tax.\n",
|
||
"23. Overdose prevention, treatment, and outreach programs funded by settlements with opioid manufacturers and distributors.\n",
|
||
"24. Savings targets in FY 2024-25 by reducing budgets for real estate expenses, software and technology licenses, and materials and supplies.\n",
|
||
"25. Lower-than-planned investments in citywide equipment, IT, and capital spending.\n",
|
||
"26. Use of reserves to help balance the budget while maintaining the bulk of the City’s reserves.\n"
|
||
]
|
||
}
|
||
],
|
||
"source": [
|
||
"response = base_query_engine.query(\n",
|
||
" \"Give me all the programs that the mayor's budget includes\"\n",
|
||
")\n",
|
||
"print(str(response))"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "58dee948-240c-4523-9be1-4f770f549a7a",
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"print(len(response.source_nodes))\n",
|
||
"for node in response.source_nodes:\n",
|
||
" print(\"---\")\n",
|
||
" print(node.get_content())"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"id": "cfabc20e-bc8b-4e80-80ef-cfdee07a6e10",
|
||
"metadata": {},
|
||
"source": [
|
||
"## Build Custom Retriever\n",
|
||
"\n",
|
||
"Build joint retriever that combines vector and KG search."
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "ea75c09f-a0d1-48fe-b349-5e52ffe4df03",
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"from llama_index.core.retrievers import BaseRetriever\n",
|
||
"from llama_index.core.schema import NodeWithScore\n",
|
||
"from typing import List\n",
|
||
"\n",
|
||
"\n",
|
||
"class CustomRetriever(BaseRetriever):\n",
|
||
" \"\"\"Custom retriever that performs both KG vector search and direct vector search.\"\"\"\n",
|
||
"\n",
|
||
" def __init__(self, kg_retriever, vector_retriever):\n",
|
||
" self._kg_retriever = kg_retriever\n",
|
||
" self._vector_retriever = vector_retriever\n",
|
||
"\n",
|
||
" def _retrieve(self, query_bundle) -> List[NodeWithScore]:\n",
|
||
" \"\"\"Retrieve nodes given query.\"\"\"\n",
|
||
" kg_nodes = self._kg_retriever.retrieve(query_bundle)\n",
|
||
" vector_nodes = self._vector_retriever.retrieve(query_bundle)\n",
|
||
"\n",
|
||
" unique_nodes = {n.node_id: n for n in kg_nodes}\n",
|
||
" unique_nodes.update({n.node_id: n for n in vector_nodes})\n",
|
||
" return list(unique_nodes.values())"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "b3dae1e1-b3a3-4896-91e5-1de95ed32a0d",
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"custom_retriever = CustomRetriever(kg_retriever, base_retriever)"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "c6c71c51-32a5-4420-917c-6d42b33c2c64",
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"nodes = custom_retriever.retrieve(\n",
|
||
" \"Give me all the programs that the mayor's budget includes\"\n",
|
||
")\n",
|
||
"# len(nodes)"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"id": "389f6d0c-12b2-45e3-916a-f521befd6b91",
|
||
"metadata": {},
|
||
"source": [
|
||
"## Build Agent\n",
|
||
"\n",
|
||
"Now that we have the retriever, we can treat it as a RAG pipeline tool, and wrap it with an agent that can perform basic CoT reasoning and maintain conversation memory over time."
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "3f7e7854-5d4e-49b8-baaa-8ad1cba053a0",
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"from llama_index.core.tools import QueryEngineTool, ToolMetadata\n",
|
||
"from llama_index.core.query_engine import RetrieverQueryEngine\n",
|
||
"\n",
|
||
"kg_query_engine = RetrieverQueryEngine(custom_retriever)\n",
|
||
"kg_query_tool = QueryEngineTool(\n",
|
||
" query_engine=kg_query_engine,\n",
|
||
" metadata=ToolMetadata(\n",
|
||
" name=\"query_tool\",\n",
|
||
" description=\"Provides information about the 2023 SF Budget Report.\",\n",
|
||
" ),\n",
|
||
")"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "ec7e8a74-2e9a-4e20-91a7-d997160af829",
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"from llama_index.core.agent import FunctionCallingAgentWorker\n",
|
||
"\n",
|
||
"agent_worker = FunctionCallingAgentWorker.from_tools(\n",
|
||
" [kg_query_tool],\n",
|
||
" llm=llm,\n",
|
||
" verbose=True,\n",
|
||
" allow_parallel_tool_calls=False,\n",
|
||
")\n",
|
||
"agent = agent_worker.as_agent()"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"id": "d567f1e8-621d-4107-9e8f-75051c900ff3",
|
||
"metadata": {},
|
||
"source": [
|
||
"## Try out Queries\n",
|
||
"\n",
|
||
"Now that the agent is setup, let's try out some queries."
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "2d0b3074-92bf-4e0e-84da-8c29d2eb3ac0",
|
||
"metadata": {},
|
||
"outputs": [
|
||
{
|
||
"name": "stdout",
|
||
"output_type": "stream",
|
||
"text": [
|
||
"Added user message to memory: Give me all the programs that the mayor's budget includes\n",
|
||
"=== Calling Function ===\n",
|
||
"Calling function: query_tool with args: {\"input\": \"all programs included in the mayor's budget\"}\n",
|
||
"=== Function Output ===\n",
|
||
"The mayor's budget includes a variety of programs and investments:\n",
|
||
"\n",
|
||
"1. **CARE Court Implementation**: Funding for engagement and assessment staff, new City attorneys, increased capacity for treatment and housing, and outreach and educational efforts.\n",
|
||
"2. **Laguna Honda Hospital**: Over $3.5 million for staffing in key areas, including education and training, patient care experience, medication management, and leadership.\n",
|
||
"3. **Economic Recovery**: $24.4 million over two years for the Roadmap for Downtown San Francisco’s Future and economic recovery across the city.\n",
|
||
"4. **Tax Relief and Incentives**: Changes to business taxes, including delaying tax increases for certain industries and offering discounts on office-based gross receipts tax for new offices.\n",
|
||
"5. **Small Business Support**: $5 million in direct grants to help small businesses stabilize, scale, and adapt.\n",
|
||
"6. **CalAIM Expansion**: Focus on expanded benefits for people at risk of long-term institutionalization and justice-system involved individuals.\n",
|
||
"7. **Mental Health Services Act**: $32.0 million in FY 2022-23 and $17.7 million ongoing for mental health services, including a $15.0 million pilot for a talk therapy telehealth program.\n",
|
||
"8. **Fee for Service Transition**: Expansion of the fee-for-service model to maximize revenues.\n",
|
||
"9. **Street Overdose Response Team (SORT)**: Efforts to reduce opioid-related deaths and provide support to overdose survivors.\n",
|
||
"10. **Capital and IT Investments**: $118 million over two years for urgent repairs and $53.9 million for vital technology projects.\n",
|
||
"11. **Dream Keeper Initiative**: Continued investment in San Francisco’s Black communities, including support for small businesses, housing programs, and health services.\n",
|
||
"12. **Climate Action Plan**: $2 million over two years to support staff at the Department of Environment.\n",
|
||
"13. **Good Government Initiatives**: Funding for hiring and contracting reforms, ongoing and new IT projects, and capital maintenance and critical repairs.\n",
|
||
"14. **Minimum Compensation Ordinance (MCO)**: Investments to increase wages for the lowest-paid workers providing City services.\n",
|
||
"\n",
|
||
"These programs reflect a broad range of priorities, including health, economic recovery, small business support, mental health services, and infrastructure improvements.\n",
|
||
"=== LLM Response ===\n",
|
||
"The mayor's budget includes a comprehensive array of programs and investments, covering various sectors and priorities:\n",
|
||
"\n",
|
||
"1. **CARE Court Implementation**: Funding for engagement and assessment staff, new City attorneys, increased capacity for treatment and housing, and outreach and educational efforts.\n",
|
||
"2. **Laguna Honda Hospital**: Over $3.5 million for staffing in key areas, including education and training, patient care experience, medication management, and leadership.\n",
|
||
"3. **Economic Recovery**: $24.4 million over two years for the Roadmap for Downtown San Francisco’s Future and economic recovery across the city.\n",
|
||
"4. **Tax Relief and Incentives**: Changes to business taxes, including delaying tax increases for certain industries and offering discounts on office-based gross receipts tax for new offices.\n",
|
||
"5. **Small Business Support**: $5 million in direct grants to help small businesses stabilize, scale, and adapt.\n",
|
||
"6. **CalAIM Expansion**: Focus on expanded benefits for people at risk of long-term institutionalization and justice-system involved individuals.\n",
|
||
"7. **Mental Health Services Act**: $32.0 million in FY 2022-23 and $17.7 million ongoing for mental health services, including a $15.0 million pilot for a talk therapy telehealth program.\n",
|
||
"8. **Fee for Service Transition**: Expansion of the fee-for-service model to maximize revenues.\n",
|
||
"9. **Street Overdose Response Team (SORT)**: Efforts to reduce opioid-related deaths and provide support to overdose survivors.\n",
|
||
"10. **Capital and IT Investments**: $118 million over two years for urgent repairs and $53.9 million for vital technology projects.\n",
|
||
"11. **Dream Keeper Initiative**: Continued investment in San Francisco’s Black communities, including support for small businesses, housing programs, and health services.\n",
|
||
"12. **Climate Action Plan**: $2 million over two years to support staff at the Department of Environment.\n",
|
||
"13. **Good Government Initiatives**: Funding for hiring and contracting reforms, ongoing and new IT projects, and capital maintenance and critical repairs.\n",
|
||
"14. **Minimum Compensation Ordinance (MCO)**: Investments to increase wages for the lowest-paid workers providing City services.\n",
|
||
"\n",
|
||
"These programs reflect a broad range of priorities, including health, economic recovery, small business support, mental health services, and infrastructure improvements.\n"
|
||
]
|
||
}
|
||
],
|
||
"source": [
|
||
"response = agent.chat(\"Give me all the programs that the mayor's budget includes\")"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "f1d0c63e-817f-4615-a237-82d728917e4e",
|
||
"metadata": {},
|
||
"outputs": [
|
||
{
|
||
"name": "stdout",
|
||
"output_type": "stream",
|
||
"text": [
|
||
"The mayor's budget includes a comprehensive array of programs and investments, covering various sectors and priorities:\n",
|
||
"\n",
|
||
"1. **CARE Court Implementation**: Funding for engagement and assessment staff, new City attorneys, increased capacity for treatment and housing, and outreach and educational efforts.\n",
|
||
"2. **Laguna Honda Hospital**: Over $3.5 million for staffing in key areas, including education and training, patient care experience, medication management, and leadership.\n",
|
||
"3. **Economic Recovery**: $24.4 million over two years for the Roadmap for Downtown San Francisco’s Future and economic recovery across the city.\n",
|
||
"4. **Tax Relief and Incentives**: Changes to business taxes, including delaying tax increases for certain industries and offering discounts on office-based gross receipts tax for new offices.\n",
|
||
"5. **Small Business Support**: $5 million in direct grants to help small businesses stabilize, scale, and adapt.\n",
|
||
"6. **CalAIM Expansion**: Focus on expanded benefits for people at risk of long-term institutionalization and justice-system involved individuals.\n",
|
||
"7. **Mental Health Services Act**: $32.0 million in FY 2022-23 and $17.7 million ongoing for mental health services, including a $15.0 million pilot for a talk therapy telehealth program.\n",
|
||
"8. **Fee for Service Transition**: Expansion of the fee-for-service model to maximize revenues.\n",
|
||
"9. **Street Overdose Response Team (SORT)**: Efforts to reduce opioid-related deaths and provide support to overdose survivors.\n",
|
||
"10. **Capital and IT Investments**: $118 million over two years for urgent repairs and $53.9 million for vital technology projects.\n",
|
||
"11. **Dream Keeper Initiative**: Continued investment in San Francisco’s Black communities, including support for small businesses, housing programs, and health services.\n",
|
||
"12. **Climate Action Plan**: $2 million over two years to support staff at the Department of Environment.\n",
|
||
"13. **Good Government Initiatives**: Funding for hiring and contracting reforms, ongoing and new IT projects, and capital maintenance and critical repairs.\n",
|
||
"14. **Minimum Compensation Ordinance (MCO)**: Investments to increase wages for the lowest-paid workers providing City services.\n",
|
||
"\n",
|
||
"These programs reflect a broad range of priorities, including health, economic recovery, small business support, mental health services, and infrastructure improvements.\n"
|
||
]
|
||
}
|
||
],
|
||
"source": [
|
||
"print(str(response))"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "a7cb264f-70c8-4d02-a11b-2c8b8990af76",
|
||
"metadata": {},
|
||
"outputs": [
|
||
{
|
||
"name": "stdout",
|
||
"output_type": "stream",
|
||
"text": [
|
||
"Added user message to memory: Compare the budget for DPA Police Accountabilty from 2022-2023 to 2023-2024\n",
|
||
"=== Calling Function ===\n",
|
||
"Calling function: query_tool with args: {\"input\": \"DPA Police Accountability budget for 2022-2023\"}\n",
|
||
"=== Function Output ===\n",
|
||
"The DPA Police Accountability budget for 2022-2023 is $9,776,177.\n",
|
||
"=== Calling Function ===\n",
|
||
"Calling function: query_tool with args: {\"input\": \"DPA Police Accountability budget for 2023-2024\"}\n",
|
||
"=== Function Output ===\n",
|
||
"The budget for the Department of Police Accountability for the fiscal year 2023-2024 is $9,990,353.\n",
|
||
"=== LLM Response ===\n",
|
||
"The budget for the Department of Police Accountability (DPA) has increased from $9,776,177 in the fiscal year 2022-2023 to $9,990,353 in the fiscal year 2023-2024. This represents an increase of $214,176.\n",
|
||
"The budget for the Department of Police Accountability (DPA) has increased from $9,776,177 in the fiscal year 2022-2023 to $9,990,353 in the fiscal year 2023-2024. This represents an increase of $214,176.\n"
|
||
]
|
||
}
|
||
],
|
||
"source": [
|
||
"agent.reset()\n",
|
||
"response = agent.chat(\n",
|
||
" \"Compare the budget for DPA Police Accountabilty from 2022-2023 to 2023-2024\"\n",
|
||
")\n",
|
||
"print(str(response))"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "4ece9079-da9c-4573-b249-2cab9d71fddf",
|
||
"metadata": {},
|
||
"outputs": [
|
||
{
|
||
"name": "stdout",
|
||
"output_type": "stream",
|
||
"text": [
|
||
"Here are some facts extracted from the provided text:\n",
|
||
"\n",
|
||
"Dpa department of police accountability -> Has -> Total funded positions\n",
|
||
"\n",
|
||
"ORGANIZATIONAL STRUCTURE: POLICE ACCOUNTABILITY\n",
|
||
"\n",
|
||
"\n",
|
||
" Executive Director\n",
|
||
"\n",
|
||
"\n",
|
||
" Chief of Staff Chief of Investigations\n",
|
||
"\n",
|
||
"\n",
|
||
" Audit Operations Legal Investigations Mediation\n",
|
||
"\n",
|
||
"\n",
|
||
" Policy SB 1421\n",
|
||
"\n",
|
||
"\n",
|
||
"Department Total Budget Historical Comparison (Mayor's Proposed) Budget Year 2023-2024 and 2024-2025\n",
|
||
"\n",
|
||
"\n",
|
||
" Department Total Budget Historical Comparison\n",
|
||
" DPA Department Of Police AccountabilityTOTAL BUDGET – HISTORICAL COMPARISON2022-2023 2023-2024 CHANGE 2024-2025 CHANGE\n",
|
||
" FUNDED POSITIONS ORIGINAL PROPOSED FROM PROPOSED FROM\n",
|
||
" 2022-2023BUDGET 2023-2024BUDGETChanges from2022-2023 2024-2025BUDGETChanges from2023-2024\n",
|
||
"Funded Positions Original Budget Proposed Budget 2022-2023 Proposed Budget 2023-2024\n",
|
||
" Total Funded 45.17 41.95 (3.22) 41.85 (0.10)\n",
|
||
" Non-Operating Positions (CAP/Other) (2.00) (1.00) 1.00 (1.00)\n",
|
||
" Net Operating Positions 43.17 40.95 (2.22) 40.85 (0.10)\n",
|
||
"\n",
|
||
"\n",
|
||
"Sources\n",
|
||
" Expenditure Recovery 128,000 332,795 204,795 332,795\n",
|
||
"\n",
|
||
"\n",
|
||
" General Fund 9,648,177 9,657,558 9,381 9,488,396 (169,162)\n",
|
||
" Sources Total 9,776,177 9,990,353 214,176 9,821,191 (169,162)\n",
|
||
"\n",
|
||
"\n",
|
||
"Uses - Operating Expenditures\n",
|
||
" Salaries 6,003,750 5,930,159 (73,591) 6,158,974 228,815\n",
|
||
" Mandatory Fringe Benefits 2,257,157 2,086,784 (170,373) 2,143,678 56,894\n",
|
||
" Non-Personnel Services 324,336 334,336 10,000 333,742 (594)\n",
|
||
" Materials & Supplies 34,918 34,918 31,426 (3,492)\n",
|
||
" Programmatic Projects 100,000 500,000 400,000 100,000 (400,000)\n",
|
||
" Services Of Other Depts 1,056,016 1,104,156 48,140 1,053,371 (50,785)\n",
|
||
" Uses Total 9,776,177 9,990,353 214,176 9,821,191 (169,162)\n",
|
||
"\n",
|
||
"\n",
|
||
"Uses - By Division Description\n",
|
||
" DPA Police Accountabilty 9,776,177 9,990,353 214,176 9,821,191 (169,162)\n",
|
||
" Uses by Division Total 9,776,177 9,990,353 214,176 9,821,191 (169,162)\n",
|
||
"\n",
|
||
"\n",
|
||
" POLICE ACCOUNTABILITY 249\n"
|
||
]
|
||
}
|
||
],
|
||
"source": [
|
||
"print(str(response.source_nodes[0].get_content()))"
|
||
]
|
||
}
|
||
],
|
||
"metadata": {
|
||
"kernelspec": {
|
||
"display_name": "llama_parse",
|
||
"language": "python",
|
||
"name": "llama_parse"
|
||
},
|
||
"language_info": {
|
||
"codemirror_mode": {
|
||
"name": "ipython",
|
||
"version": 3
|
||
},
|
||
"file_extension": ".py",
|
||
"mimetype": "text/x-python",
|
||
"name": "python",
|
||
"nbconvert_exporter": "python",
|
||
"pygments_lexer": "ipython3"
|
||
}
|
||
},
|
||
"nbformat": 4,
|
||
"nbformat_minor": 5
|
||
}
|