├── .env.example ├── .gitignore ├── Makefile ├── README.md ├── basic-diagrams.excalidraw ├── notebooks ├── 1.0-intro-to-langchain.ipynb ├── 1.1-intro-to-runnable-interface.ipynb ├── 2.0-qa-with-langchain.ipynb ├── 2.1-langchain-query-csv.ipynb ├── 2.2-dynamic-quiz-over-pdf.ipynb ├── 2.3-qa-fully-local.ipynb ├── 2.4-simple-rag-langchain-langgraph.ipynb ├── 3.0-building-llm-agents-with-langchain.ipynb ├── 3.1-langchain-github-agent-prototype.ipynb ├── 3.2-langchain-agent-with-langgraph.ipynb ├── 4.0-langgraph-quick-introduction.ipynb ├── 5.0-demos-research-workflows.ipynb ├── 6.0-live-demo-chat-with-langchain-docs-urls.ipynb ├── README.md ├── assets-resources │ ├── Augmenting-Human-Intellect_ A-Conceptual-Framework-1962-Doug Engelbart.pdf │ ├── agent-loop-langgraph-version.png │ ├── agent-loop.png │ ├── agent_loop.svg │ ├── ai.png │ ├── attention-paper.pdf │ ├── attributes_runnable.png │ ├── basic-agent-diagram.png │ ├── chatgpt-demo.png │ ├── components_input_output.png │ ├── components_input_type_output_type.png │ ├── diagram_langchain_agent_find_source_code.png │ ├── embeddings-similarity.png │ ├── embeddings-similarity2.png │ ├── embeddings.png │ ├── google-icon.png │ ├── google-translate.png │ ├── images_tbn.jpg │ ├── langchain-icon.svg │ ├── langchain-project-structure.png │ ├── langchain-toolkits.png │ ├── langgraph-components.png │ ├── lcel-image.png │ ├── llm_paper_know_dont_know.pdf │ ├── llm_predicts_pancakes.png │ ├── pancake_maker.png │ ├── paper-llm-components.pdf │ ├── prompts.csv │ ├── pydantic.png │ ├── rag-docs.png │ ├── rag-langchain-retrieval.png │ ├── rag-langchain.png │ ├── superheroes.csv │ ├── tagging.png │ └── vectordb.png ├── dev-notebooks │ ├── 2023-12-13-14-12-45.png │ ├── 2023-12-13-14-25-37.png │ ├── 5.1-demo-playwright-scrape-articles.py │ ├── 6.0-langchain-deploy-recipe-walkthrough.ipynb │ ├── 6.1-langchain-app-dev-structure.ipynb │ ├── 7.0-fun-demos-adept.ipynb │ ├── create-knowledge-graph.ipynb │ ├── knowledge_graph.gv │ ├── knowledge_graph.gv.pdf │ ├── live-coding-demo-agents-student-request.ipynb │ ├── live-coding-demo-agents.ipynb │ ├── live-coding-demo-qa-pdf.ipynb │ ├── llama3-langchain-basic-chain.ipynb │ └── research-assistant_test.py ├── intro-rag-basics.ipynb ├── jira-agent.py ├── langchain-app.py ├── langchain-lcel-cheatsheet.md ├── langchain-structured-output-ui.py ├── live-example-liner-regression-langgraph.ipynb ├── live-notebook-rag-basics.ipynb ├── live-session-intro.ipynb ├── paper.pdf ├── rag_methods.py ├── research-agent-langgraph.ipynb ├── simple-product-info-chatbot.py └── testing-langchain-app.ipynb ├── presentation_slides ├── 2024-02-20-21-43-40.png ├── Getting-Started-with-LangChain-presentation.pdf ├── extract-presentation.applescript ├── presentation-langgraph.html └── presentation.html └── requirements ├── requirements.in └── requirements.txt /.env.example: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/.env.example -------------------------------------------------------------------------------- /.gitignore: -------------------------------------------------------------------------------- 1 | # Ignore node_modules directory 2 | node_modules/ 3 | 4 | # Ignore build artifacts 5 | dist/ 6 | build/ 7 | 8 | # Ignore environment-specific files 9 | .env 10 | .env.local 11 | .env.*.local 12 | 13 | # Ignore log files 14 | *.log 15 | *.sqlite 16 | *.bin 17 | 18 | # Ignore editor-specific files 19 | .vscode/ 20 | .idea/ 21 | 22 | # Ignore OS-specific files 23 | .DS_Store 24 | Thumbs.db 25 | 26 | # Project files 27 | 28 | ## presentation-slides folder 29 | presentation-slides/ 30 | 31 | -------------------------------------------------------------------------------- /Makefile: -------------------------------------------------------------------------------- 1 | ENV_NAME ?= oreilly-langchain 2 | PYTHON_VERSION ?= 3.11 3 | 4 | # Install exact Python and CUDA versions 5 | conda-create: 6 | conda create -n $(ENV_NAME) python=$(PYTHON_VERSION) 7 | 8 | repo-setup: 9 | mkdir requirements 10 | touch requirements/requirements.in 11 | pip install uv 12 | uv pip install pip-tools 13 | 14 | notebook-setup: 15 | python -m ipykernel install --user --name=$(ENV_NAME) 16 | 17 | env-update: 18 | uv pip compile ./requirements/requirements.in -o ./requirements/requirements.txt 19 | uv pip sync ./requirements/requirements.txt 20 | 21 | pip-tools-setup: 22 | pip install uv 23 | uv pip install pip-tools setuptools 24 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # Notebooks for the O'Reilly live-training: "Getting Started with LangChain" 2 | 3 | - [Live-training official website from O'Reilly](https://learning.oreilly.com/live-events/getting-started-with-langchain/0636920098586/0636920098585/) 4 | # Overview 5 | 6 | 7 | ## Notebooks 8 | 9 | 1. **Getting Started with LangChain** 10 | - [Introduction to LangChain](https://colab.research.google.com/github/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/blob/main/notebooks/1.0-intro-to-langchain.ipynb) 11 | 12 | 2. **Advanced Query and Dynamic Content** 13 | - [QA with LangChain](https://colab.research.google.com/github/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/blob/main/notebooks/2.0-qa-with-langchain.ipynb) 14 | - [Querying CSV Data with LangChain](https://colab.research.google.com/github/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/blob/main/notebooks/2.1-langchain-query-csv.ipynb) 15 | - [Dynamic Quiz over PDF](https://colab.research.google.com/github/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/blob/main/notebooks/2.2-dynamic-quiz-over-pdf.ipynb) 16 | - [QA Fully Local](https://colab.research.google.com/github/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/blob/main/notebooks/2.3-qa-fully-local.ipynb) 17 | 18 | 3. **Building Intelligent Agents** 19 | - [Building LLM Agents with LangChain](https://colab.research.google.com/github/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/blob/main/notebooks/3.0-building-llm-agents-with-langchain.ipynb) 20 | - [LangChain GitHub Agent Prototype](https://colab.research.google.com/github/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/blob/main/notebooks/3.1-langchain-github-agent-prototype.ipynb) 21 | 22 | 4. **Quick Introduction to LangGraph** 23 | - [Quick Introduction to LangGraph](https://colab.research.google.com/github/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/blob/main/notebooks/4.0-langgraph-quick-introduction.ipynb) 24 | 25 | 5. **Demonstrations and Practical Applications** 26 | - [Research Workflows](https://colab.research.google.com/github/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/blob/main/notebooks/5.0-demos-research-workflows.ipynb) 27 | 28 | 6. **Additional Resources** 29 | - [Live Coding Demo QA PDF](https://colab.research.google.com/github/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/blob/main/notebooks/live-coding-demo-qa-pdf.ipynb) 30 | 31 | 32 | ## Setup 33 | 34 | **Conda** 35 | 36 | - Install [anaconda](https://www.anaconda.com/download) 37 | - Create an environment: `conda create -n oreilly-langchain python=3.11` 38 | - Activate your environment with: `conda activate oreilly-langchain` 39 | - Install requirements with: `pip install -r requirements.txt` 40 | - Setup your openai [API key](https://platform.openai.com/) 41 | 42 | **If you're used to Makefiles** 43 | - Install `uv` with `pip install uv` 44 | - Run `make all` 45 | - Activate your environment just like below and you're done 46 | 47 | **Pip** 48 | 49 | 1. **Create a Virtual Environment:** 50 | Navigate to your project directory. If using Python 3's built-in `venv`: 51 | ```bash 52 | python -m venv oreilly_env 53 | ``` 54 | If you're using `virtualenv`: 55 | ```bash 56 | virtualenv oreilly_env 57 | ``` 58 | 59 | 2. **Activate the Virtual Environment:** 60 | - **On Windows:** 61 | ```bash 62 | .\oreilly_env\Scripts\activate 63 | ``` 64 | - **On macOS and Linux:** 65 | ```bash 66 | source oreilly_env/bin/activate 67 | ``` 68 | 69 | 3. **Install Dependencies from `requirements.txt`:** 70 | ```bash 71 | pip install python-dotenv 72 | pip install -r requirements.txt 73 | ``` 74 | 75 | 4. Setup your openai [API key](https://platform.openai.com/) 76 | 77 | Remember to deactivate the virtual environment once you're done by simply typing: 78 | ```bash 79 | deactivate 80 | ``` 81 | 82 | ## Setup your .env file 83 | 84 | - Change the `.env.example` file to `.env` and add your OpenAI API key. 85 | 86 | ## To use this Environment with Jupyter Notebooks: 87 | 88 | ```python3 -m ipykernel install --user --name=oreilly-langchain``` 89 | 90 | ## Official Training Website 91 | 92 | For more information about the live-training, visit the [official website](https://learning.oreilly.com/live-events/getting-started-with-langchain/0636920098586/0636920098585/). 93 | -------------------------------------------------------------------------------- /notebooks/2.1-langchain-query-csv.ipynb: -------------------------------------------------------------------------------- 1 | { 2 | "cells": [ 3 | { 4 | "cell_type": "code", 5 | "execution_count": 4, 6 | "metadata": {}, 7 | "outputs": [], 8 | "source": [ 9 | "# %pip install langchain\n", 10 | "# # %pip install langchain-openai\n", 11 | "# # %pip install langchainhub\n", 12 | "# # %pip install pypdf\n", 13 | "# %pip install chromadb\n", 14 | "# %pip install pandas" 15 | ] 16 | }, 17 | { 18 | "cell_type": "markdown", 19 | "metadata": {}, 20 | "source": [ 21 | "# DISCLAIMER: \n", 22 | "\n", 23 | "```\n", 24 | "FOR THIS NOTEBOOK YOU WILL NEED TO BE RUNNING LOCALLY OLLAMA + HAVE THE LLAMA 3.1 MODEL INSTALLED\n", 25 | "```\n", 26 | "\n", 27 | "See [here](https://python.langchain.com/v0.2/docs/integrations/chat/ollama/) for how to set that up." 28 | ] 29 | }, 30 | { 31 | "cell_type": "code", 32 | "execution_count": 1, 33 | "metadata": {}, 34 | "outputs": [], 35 | "source": [ 36 | "import os\n", 37 | "import getpass\n", 38 | "\n", 39 | "# Set OPENAI API Key\n", 40 | "\n", 41 | "import os\n", 42 | "import getpass\n", 43 | "\n", 44 | "def _set_env(var: str):\n", 45 | " if not os.environ.get(var):\n", 46 | " os.environ[var] = getpass.getpass(f\"var: \")\n", 47 | "\n", 48 | "_set_env(\"OPENAI_API_KEY\")" 49 | ] 50 | }, 51 | { 52 | "cell_type": "markdown", 53 | "metadata": {}, 54 | "source": [ 55 | "A vector database is a way to store these embeddings, these numerical representations that we just discussed.\n", 56 | "\n", 57 | "The pipeline is:\n", 58 | "- In coming document\n", 59 | "- Create chunks of text from that document\n", 60 | "- Embed each chunk\n", 61 | "- Store these embeddings\n", 62 | "\n", 63 | "![](2023-07-30-19-32-13.png)\n", 64 | "\n", 65 | "[LangChain for LLM Application Development by Deeplearning.ai](https://learn.deeplearning.ai/langchain/lesson/1/introduction)" 66 | ] 67 | }, 68 | { 69 | "cell_type": "code", 70 | "execution_count": 2, 71 | "metadata": {}, 72 | "outputs": [ 73 | { 74 | "data": { 75 | "text/html": [ 76 | "
\n", 77 | "\n", 90 | "\n", 91 | " \n", 92 | " \n", 93 | " \n", 94 | " \n", 95 | " \n", 96 | " \n", 97 | " \n", 98 | " \n", 99 | " \n", 100 | " \n", 101 | " \n", 102 | " \n", 103 | " \n", 104 | " \n", 105 | " \n", 106 | " \n", 107 | " \n", 108 | " \n", 109 | " \n", 110 | " \n", 111 | " \n", 112 | " \n", 113 | " \n", 114 | " \n", 115 | " \n", 116 | " \n", 117 | " \n", 118 | " \n", 119 | " \n", 120 | " \n", 121 | " \n", 122 | " \n", 123 | " \n", 124 | " \n", 125 | " \n", 126 | " \n", 127 | " \n", 128 | " \n", 129 | " \n", 130 | " \n", 131 | " \n", 132 | " \n", 133 | " \n", 134 | " \n", 135 | " \n", 136 | " \n", 137 | "
Superhero NameSuperpowerPower LevelCatchphrase
0Captain ThunderBolt Manipulation90Feel the power of the storm!
1Silver FalconFlight and Agility85Soar high, fearlessly!
2Mystic ShadowInvisibility and Illusions78Disappear into the darkness!
3Blaze RunnerPyrokinesis88Burn bright and fierce!
4Electra-WaveElectric Manipulation82Unleash the electric waves!
\n", 138 | "
" 139 | ], 140 | "text/plain": [ 141 | " Superhero Name Superpower Power Level \\\n", 142 | "0 Captain Thunder Bolt Manipulation 90 \n", 143 | "1 Silver Falcon Flight and Agility 85 \n", 144 | "2 Mystic Shadow Invisibility and Illusions 78 \n", 145 | "3 Blaze Runner Pyrokinesis 88 \n", 146 | "4 Electra-Wave Electric Manipulation 82 \n", 147 | "\n", 148 | " Catchphrase \n", 149 | "0 Feel the power of the storm! \n", 150 | "1 Soar high, fearlessly! \n", 151 | "2 Disappear into the darkness! \n", 152 | "3 Burn bright and fierce! \n", 153 | "4 Unleash the electric waves! " 154 | ] 155 | }, 156 | "execution_count": 2, 157 | "metadata": {}, 158 | "output_type": "execute_result" 159 | } 160 | ], 161 | "source": [ 162 | "import pandas as pd\n", 163 | "\n", 164 | "file_path=\"./assets-resources/superheroes.csv\"\n", 165 | "df = pd.read_csv(file_path)\n", 166 | "\n", 167 | "df.head()" 168 | ] 169 | }, 170 | { 171 | "cell_type": "code", 172 | "execution_count": 3, 173 | "metadata": {}, 174 | "outputs": [], 175 | "source": [ 176 | "from langchain.document_loaders.csv_loader import CSVLoader" 177 | ] 178 | }, 179 | { 180 | "cell_type": "code", 181 | "execution_count": 4, 182 | "metadata": {}, 183 | "outputs": [ 184 | { 185 | "data": { 186 | "text/plain": [ 187 | "[Document(metadata={'source': './assets-resources/superheroes.csv', 'row': 0}, page_content='Superhero Name: Captain Thunder\\nSuperpower: Bolt Manipulation\\nPower Level: 90\\nCatchphrase: Feel the power of the storm!'),\n", 188 | " Document(metadata={'source': './assets-resources/superheroes.csv', 'row': 1}, page_content='Superhero Name: Silver Falcon\\nSuperpower: Flight and Agility\\nPower Level: 85\\nCatchphrase: Soar high, fearlessly!'),\n", 189 | " Document(metadata={'source': './assets-resources/superheroes.csv', 'row': 2}, page_content='Superhero Name: Mystic Shadow\\nSuperpower: Invisibility and Illusions\\nPower Level: 78\\nCatchphrase: Disappear into the darkness!'),\n", 190 | " Document(metadata={'source': './assets-resources/superheroes.csv', 'row': 3}, page_content='Superhero Name: Blaze Runner\\nSuperpower: Pyrokinesis\\nPower Level: 88\\nCatchphrase: Burn bright and fierce!'),\n", 191 | " Document(metadata={'source': './assets-resources/superheroes.csv', 'row': 4}, page_content='Superhero Name: Electra-Wave\\nSuperpower: Electric Manipulation\\nPower Level: 82\\nCatchphrase: Unleash the electric waves!')]" 192 | ] 193 | }, 194 | "execution_count": 4, 195 | "metadata": {}, 196 | "output_type": "execute_result" 197 | } 198 | ], 199 | "source": [ 200 | "loader = CSVLoader(file_path)\n", 201 | "data = loader.load()\n", 202 | "data[:5]" 203 | ] 204 | }, 205 | { 206 | "cell_type": "code", 207 | "execution_count": 5, 208 | "metadata": {}, 209 | "outputs": [], 210 | "source": [ 211 | "from langchain_ollama import ChatOllama\n", 212 | "\n", 213 | "llm = ChatOllama(\n", 214 | " model=\"gemma3\",\n", 215 | " verbose=True,\n", 216 | ")" 217 | ] 218 | }, 219 | { 220 | "cell_type": "code", 221 | "execution_count": 6, 222 | "metadata": {}, 223 | "outputs": [], 224 | "source": [ 225 | "from langchain.text_splitter import RecursiveCharacterTextSplitter\n", 226 | "\n", 227 | "text_splitter = RecursiveCharacterTextSplitter(chunk_size=1500, chunk_overlap=100)\n", 228 | "all_splits = text_splitter.split_documents(data)" 229 | ] 230 | }, 231 | { 232 | "cell_type": "code", 233 | "execution_count": 7, 234 | "metadata": {}, 235 | "outputs": [], 236 | "source": [ 237 | "# Embed and store\n", 238 | "from langchain_chroma import Chroma\n", 239 | "from langchain_ollama import OllamaEmbeddings\n", 240 | "\n", 241 | "vectordb = Chroma.from_documents(documents=all_splits, embedding=OllamaEmbeddings(model='nomic-embed-text'))" 242 | ] 243 | }, 244 | { 245 | "cell_type": "code", 246 | "execution_count": 8, 247 | "metadata": {}, 248 | "outputs": [ 249 | { 250 | "data": { 251 | "text/plain": [ 252 | "4" 253 | ] 254 | }, 255 | "execution_count": 8, 256 | "metadata": {}, 257 | "output_type": "execute_result" 258 | } 259 | ], 260 | "source": [ 261 | "question = \"What is the name of the thunder super hero?\"\n", 262 | "docs = vectordb.similarity_search(question)\n", 263 | "len(docs)" 264 | ] 265 | }, 266 | { 267 | "cell_type": "code", 268 | "execution_count": 9, 269 | "metadata": {}, 270 | "outputs": [ 271 | { 272 | "data": { 273 | "text/plain": [ 274 | "[Document(metadata={'row': 0, 'source': './assets-resources/superheroes.csv'}, page_content='Superhero Name: Captain Thunder\\nSuperpower: Bolt Manipulation\\nPower Level: 90\\nCatchphrase: Feel the power of the storm!'),\n", 275 | " Document(metadata={'row': 30, 'source': './assets-resources/superheroes.csv'}, page_content='Superhero Name: Thunderstrike\\nSuperpower: Lightning Control\\nPower Level: 91\\nCatchphrase: Electrify the battlefield!'),\n", 276 | " Document(metadata={'row': 20, 'source': './assets-resources/superheroes.csv'}, page_content=\"Superhero Name: Stormbringer\\nSuperpower: Weather Manipulation\\nPower Level: 93\\nCatchphrase: Unleash the storm's fury!\"),\n", 277 | " Document(metadata={'row': 8, 'source': './assets-resources/superheroes.csv'}, page_content='Superhero Name: Steel Titan\\nSuperpower: Super Strength and Durability\\nPower Level: 95\\nCatchphrase: Indestructible force of nature!')]" 278 | ] 279 | }, 280 | "execution_count": 9, 281 | "metadata": {}, 282 | "output_type": "execute_result" 283 | } 284 | ], 285 | "source": [ 286 | "docs" 287 | ] 288 | }, 289 | { 290 | "cell_type": "code", 291 | "execution_count": 10, 292 | "metadata": {}, 293 | "outputs": [], 294 | "source": [ 295 | "from langchain.chains.combine_documents import create_stuff_documents_chain\n", 296 | "from langchain.chains import create_retrieval_chain\n", 297 | "from langchain_core.prompts import ChatPromptTemplate\n", 298 | "from langchain_core.runnables import RunnablePassthrough\n", 299 | "from langchain_core.output_parsers import StrOutputParser\n", 300 | "\n", 301 | "def format_docs(docs):\n", 302 | " return \"\\n\\n\".join(doc.page_content for doc in docs)\n", 303 | "\n", 304 | "\n", 305 | "system_prompt = (\n", 306 | " \"You are an assistant for question-answering tasks. \"\n", 307 | " \"Use the following pieces of retrieved context to answer \"\n", 308 | " \"the question. If you don't know the answer, say that you \"\n", 309 | " \"don't know. Use three sentences maximum and keep the \"\n", 310 | " \"answer concise.\"\n", 311 | " \"\\n\\n\"\n", 312 | " \"{context}\"\n", 313 | ")\n", 314 | "\n", 315 | "prompt = ChatPromptTemplate.from_messages([\n", 316 | " ('system', system_prompt),\n", 317 | " ('human', '{input}')\n", 318 | "])\n", 319 | "\n", 320 | "\n", 321 | "rag_chain_from_docs = (\n", 322 | " {\n", 323 | " 'input': lambda x: x['input'],\n", 324 | " 'context': lambda x: format_docs(x['context']), \n", 325 | " }\n", 326 | " | prompt\n", 327 | " | llm\n", 328 | " | StrOutputParser()\n", 329 | ")\n", 330 | "\n", 331 | "retriever = vectordb.as_retriever()\n", 332 | "\n", 333 | "# passing the input query to the retriever\n", 334 | "retrieve_docs = (lambda x: x['input']) | retriever\n", 335 | "\n", 336 | "qa_chain = RunnablePassthrough.assign(context=retrieve_docs).assign(\n", 337 | " answer=rag_chain_from_docs\n", 338 | ")" 339 | ] 340 | }, 341 | { 342 | "cell_type": "code", 343 | "execution_count": 11, 344 | "metadata": {}, 345 | "outputs": [ 346 | { 347 | "data": { 348 | "text/plain": [ 349 | "{'input': 'What is the catch phrase for the super hero with the power of producing balls of fire?',\n", 350 | " 'context': [Document(metadata={'row': 24, 'source': './assets-resources/superheroes.csv'}, page_content='Superhero Name: Blazing Comet\\nSuperpower: Fireball Projection\\nPower Level: 82\\nCatchphrase: Burn brighter than a comet!'),\n", 351 | " Document(metadata={'row': 42, 'source': './assets-resources/superheroes.csv'}, page_content='Superhero Name: Solar Flare\\nSuperpower: Solar Energy Projection\\nPower Level: 85\\nCatchphrase: Feel the burning light!'),\n", 352 | " Document(metadata={'row': 3, 'source': './assets-resources/superheroes.csv'}, page_content='Superhero Name: Blaze Runner\\nSuperpower: Pyrokinesis\\nPower Level: 88\\nCatchphrase: Burn bright and fierce!'),\n", 353 | " Document(metadata={'row': 11, 'source': './assets-resources/superheroes.csv'}, page_content='Superhero Name: Starburst\\nSuperpower: Energy Projection\\nPower Level: 83\\nCatchphrase: Ignite the cosmos!')],\n", 354 | " 'answer': 'Blazing Comet’s catchphrase is “Burn brighter than a comet!” He possesses the power of fireball projection and has a power level of 82.'}" 355 | ] 356 | }, 357 | "execution_count": 11, 358 | "metadata": {}, 359 | "output_type": "execute_result" 360 | } 361 | ], 362 | "source": [ 363 | "question = \"What is the catch phrase for the super hero with the power of producing balls of fire?\"\n", 364 | "result = qa_chain.invoke({'input': question})\n", 365 | "result" 366 | ] 367 | }, 368 | { 369 | "cell_type": "code", 370 | "execution_count": 12, 371 | "metadata": {}, 372 | "outputs": [ 373 | { 374 | "data": { 375 | "text/plain": [ 376 | "Superhero Name Blazing Comet\n", 377 | "Superpower Fireball Projection\n", 378 | "Power Level 82\n", 379 | "Catchphrase Burn brighter than a comet!\n", 380 | "Name: 24, dtype: object" 381 | ] 382 | }, 383 | "execution_count": 12, 384 | "metadata": {}, 385 | "output_type": "execute_result" 386 | } 387 | ], 388 | "source": [ 389 | "df.iloc[24]" 390 | ] 391 | }, 392 | { 393 | "cell_type": "code", 394 | "execution_count": null, 395 | "metadata": {}, 396 | "outputs": [], 397 | "source": [] 398 | } 399 | ], 400 | "metadata": { 401 | "kernelspec": { 402 | "display_name": "oreilly-langchain", 403 | "language": "python", 404 | "name": "oreilly-langchain" 405 | }, 406 | "language_info": { 407 | "codemirror_mode": { 408 | "name": "ipython", 409 | "version": 3 410 | }, 411 | "file_extension": ".py", 412 | "mimetype": "text/x-python", 413 | "name": "python", 414 | "nbconvert_exporter": "python", 415 | "pygments_lexer": "ipython3", 416 | "version": "3.11.9" 417 | } 418 | }, 419 | "nbformat": 4, 420 | "nbformat_minor": 2 421 | } 422 | -------------------------------------------------------------------------------- /notebooks/3.1-langchain-github-agent-prototype.ipynb: -------------------------------------------------------------------------------- 1 | { 2 | "cells": [ 3 | { 4 | "cell_type": "code", 5 | "execution_count": 12, 6 | "metadata": {}, 7 | "outputs": [], 8 | "source": [ 9 | "%pip install -qU langchain\n", 10 | "%pip install langchain-openai\n", 11 | "%pip install langchainhub\n", 12 | "%pip install tiktoken" 13 | ] 14 | }, 15 | { 16 | "cell_type": "code", 17 | "execution_count": 17, 18 | "metadata": {}, 19 | "outputs": [], 20 | "source": [ 21 | "import os\n", 22 | "import getpass\n", 23 | "\n", 24 | "# Set OPENAI API Key\n", 25 | "\n", 26 | "import os\n", 27 | "import getpass\n", 28 | "\n", 29 | "def _set_env(var: str):\n", 30 | " if not os.environ.get(var):\n", 31 | " os.environ[var] = getpass.getpass(f\"var: \")\n", 32 | "\n", 33 | "_set_env(\"OPENAI_API_KEY\")" 34 | ] 35 | }, 36 | { 37 | "cell_type": "markdown", 38 | "metadata": {}, 39 | "source": [ 40 | "The more complex version of this would use this pygithub package as shown in here:\n", 41 | "- https://python.langchain.com/docs/integrations/toolkits/github\n", 42 | "\n", 43 | "But let's try a simpler version that just sets up a bunch of python functions to automate different github actions like add, commit and pull requests." 44 | ] 45 | }, 46 | { 47 | "cell_type": "markdown", 48 | "metadata": {}, 49 | "source": [] 50 | }, 51 | { 52 | "cell_type": "markdown", 53 | "metadata": {}, 54 | "source": [ 55 | "Let's start with a simple commit to a branch of some repository." 56 | ] 57 | }, 58 | { 59 | "cell_type": "markdown", 60 | "metadata": {}, 61 | "source": [ 62 | "First, I'll create a github repository using some simple commands." 63 | ] 64 | }, 65 | { 66 | "cell_type": "markdown", 67 | "metadata": {}, 68 | "source": [] 69 | }, 70 | { 71 | "cell_type": "code", 72 | "execution_count": 14, 73 | "metadata": {}, 74 | "outputs": [], 75 | "source": [ 76 | "# git init -b main\n", 77 | "\n", 78 | "# git add . && git commit -m \"Some commit\"\n", 79 | "\n", 80 | "# gh repo create" 81 | ] 82 | }, 83 | { 84 | "cell_type": "code", 85 | "execution_count": 1, 86 | "metadata": {}, 87 | "outputs": [], 88 | "source": [ 89 | "MODEL='gpt-4o-mini'\n", 90 | "TEMP=0.0" 91 | ] 92 | }, 93 | { 94 | "cell_type": "markdown", 95 | "metadata": {}, 96 | "source": [ 97 | "Now, let's try asking an agent to write come code and commit the resulting code to the current branch of this repository.\n", 98 | "\n", 99 | "To do that, let's give the agent the necessary tools it will need which in this case will be python functions that perform different github actions using the `subprocess` package." 100 | ] 101 | }, 102 | { 103 | "cell_type": "markdown", 104 | "metadata": {}, 105 | "source": [ 106 | "We'll follow the basic steps for building a langchain agent:\n", 107 | "\n", 108 | "# Steps for building a simple agent:\n", 109 | "- Set up the LLM\n", 110 | "- Define the tool or toolkit (list of tools)\n", 111 | "- Set up a prompt template\n", 112 | "- Connect llm with your tools\n", 113 | "- Define your agent as a dict with keys: input, and agent scratchpad \n", 114 | "- Use the Langchain LCEL language to pipe agent, prompt the llm_with_tools variable and the output parser (which can use OpenAI function calling parser)\n", 115 | "- Create the agent loop\n", 116 | "- Wrap everything into the AgentExecutor\n", 117 | "- Invoke the agent with some query input\n" 118 | ] 119 | }, 120 | { 121 | "cell_type": "code", 122 | "execution_count": 18, 123 | "metadata": {}, 124 | "outputs": [], 125 | "source": [ 126 | "# setup the llm\n", 127 | "from langchain_openai import ChatOpenAI\n", 128 | "\n", 129 | "llm = ChatOpenAI(model=MODEL, temperature=TEMP)" 130 | ] 131 | }, 132 | { 133 | "cell_type": "markdown", 134 | "metadata": {}, 135 | "source": [ 136 | "Now, let's create a tool for the LLM" 137 | ] 138 | }, 139 | { 140 | "cell_type": "code", 141 | "execution_count": 4, 142 | "metadata": {}, 143 | "outputs": [], 144 | "source": [ 145 | "import subprocess\n", 146 | "\n", 147 | "def github_commit_tool(commit_message=\"Some commit\"):\n", 148 | " subprocess.run([\"git\", \"add\", \".\"])\n", 149 | " subprocess.run([\"git\", \"commit\", \"-m\", commit_message])\n", 150 | " subprocess.run([\"git\", \"push\", \"-u\", \"origin\", \"main\"])\n", 151 | " \n", 152 | " return \"Committed to Github\"" 153 | ] 154 | }, 155 | { 156 | "cell_type": "markdown", 157 | "metadata": {}, 158 | "source": [ 159 | "Now! Before we use it with langchain, let's test it by itself." 160 | ] 161 | }, 162 | { 163 | "cell_type": "code", 164 | "execution_count": 5, 165 | "metadata": {}, 166 | "outputs": [ 167 | { 168 | "name": "stdout", 169 | "output_type": "stream", 170 | "text": [ 171 | "[main 079c67c] Testing commit for github agent demo.\n", 172 | " 2 files changed, 73 insertions(+), 113 deletions(-)\n", 173 | " create mode 100644 notebooks/2024-08-26-12-39-39.png\n", 174 | "branch 'main' set up to track 'origin/main'.\n" 175 | ] 176 | }, 177 | { 178 | "name": "stderr", 179 | "output_type": "stream", 180 | "text": [ 181 | "To https://github.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain.git\n", 182 | " 3a97192..079c67c main -> main\n" 183 | ] 184 | }, 185 | { 186 | "data": { 187 | "text/plain": [ 188 | "'Committed to Github'" 189 | ] 190 | }, 191 | "execution_count": 5, 192 | "metadata": {}, 193 | "output_type": "execute_result" 194 | } 195 | ], 196 | "source": [ 197 | "github_commit_tool(\"Testing commit for github agent demo.\")" 198 | ] 199 | }, 200 | { 201 | "cell_type": "markdown", 202 | "metadata": {}, 203 | "source": [ 204 | "Now, let's take a look at out github repository from the terminal.\n", 205 | "\n", 206 | "![](2023-10-29-20-47-54.png)" 207 | ] 208 | }, 209 | { 210 | "cell_type": "markdown", 211 | "metadata": {}, 212 | "source": [ 213 | "Nice! It looks like we are good to go with this first simple tool!" 214 | ] 215 | }, 216 | { 217 | "cell_type": "markdown", 218 | "metadata": {}, 219 | "source": [ 220 | "Let's now make it a tool for our langchain agent by adding the @tool. " 221 | ] 222 | }, 223 | { 224 | "cell_type": "code", 225 | "execution_count": 19, 226 | "metadata": {}, 227 | "outputs": [], 228 | "source": [ 229 | "from langchain.tools import tool\n", 230 | "\n", 231 | "@tool\n", 232 | "def github_commit_tool(commit_message=\"Some commit\"):\n", 233 | " \"\"\"This function uses the subprocess package to make commits to a github repo pre-defined.\"\"\"\n", 234 | " subprocess.run([\"git\", \"add\", \".\"])\n", 235 | " subprocess.run([\"git\", \"commit\", \"-m\", commit_message])\n", 236 | " subprocess.run([\"git\", \"push\", \"-u\", \"origin\", \"main\"])\n", 237 | " \n", 238 | " return \"Committed to Github\"\n", 239 | "\n", 240 | "tools = [github_commit_tool]" 241 | ] 242 | }, 243 | { 244 | "cell_type": "markdown", 245 | "metadata": {}, 246 | "source": [ 247 | "We added some documentation to our function to abide by the requirements of the tool decorator from langchain." 248 | ] 249 | }, 250 | { 251 | "cell_type": "markdown", 252 | "metadata": {}, 253 | "source": [ 254 | "Now, let's test if a simple agent can use that tool!" 255 | ] 256 | }, 257 | { 258 | "cell_type": "markdown", 259 | "metadata": {}, 260 | "source": [ 261 | "Let's start by setting up our prompt template." 262 | ] 263 | }, 264 | { 265 | "cell_type": "code", 266 | "execution_count": 1, 267 | "metadata": {}, 268 | "outputs": [ 269 | { 270 | "data": { 271 | "text/plain": [ 272 | "[SystemMessagePromptTemplate(prompt=PromptTemplate(input_variables=[], input_types={}, partial_variables={}, template='You are a helpful assistant'), additional_kwargs={}),\n", 273 | " MessagesPlaceholder(variable_name='chat_history', optional=True),\n", 274 | " HumanMessagePromptTemplate(prompt=PromptTemplate(input_variables=['input'], input_types={}, partial_variables={}, template='{input}'), additional_kwargs={}),\n", 275 | " MessagesPlaceholder(variable_name='agent_scratchpad')]" 276 | ] 277 | }, 278 | "execution_count": 1, 279 | "metadata": {}, 280 | "output_type": "execute_result" 281 | } 282 | ], 283 | "source": [ 284 | "from langchain import hub\n", 285 | "\n", 286 | "prompt = hub.pull(\"hwchase17/openai-tools-agent\")\n", 287 | "prompt.messages" 288 | ] 289 | }, 290 | { 291 | "cell_type": "code", 292 | "execution_count": 8, 293 | "metadata": {}, 294 | "outputs": [], 295 | "source": [ 296 | "from langchain.agents import create_tool_calling_agent\n", 297 | "from langchain_openai import ChatOpenAI\n", 298 | "\n", 299 | "agent = create_tool_calling_agent(llm, tools, prompt)" 300 | ] 301 | }, 302 | { 303 | "cell_type": "code", 304 | "execution_count": 9, 305 | "metadata": {}, 306 | "outputs": [], 307 | "source": [ 308 | "from langchain.agents import AgentExecutor\n", 309 | "\n", 310 | "agent_executor = AgentExecutor(agent=agent,tools=tools, verbose=True)" 311 | ] 312 | }, 313 | { 314 | "cell_type": "code", 315 | "execution_count": 10, 316 | "metadata": {}, 317 | "outputs": [ 318 | { 319 | "name": "stdout", 320 | "output_type": "stream", 321 | "text": [ 322 | "\n", 323 | "\n", 324 | "\u001b[1m> Entering new AgentExecutor chain...\u001b[0m\n", 325 | "\u001b[32;1m\u001b[1;3m\n", 326 | "Invoking: `github_commit_tool` with `{'commit_message': 'Testing LLM agent!'}`\n", 327 | "\n", 328 | "\n", 329 | "\u001b[0mOn branch main\n", 330 | "Your branch is up to date with 'origin/main'.\n", 331 | "\n", 332 | "nothing to commit, working tree clean\n" 333 | ] 334 | }, 335 | { 336 | "name": "stderr", 337 | "output_type": "stream", 338 | "text": [ 339 | "Everything up-to-date\n" 340 | ] 341 | }, 342 | { 343 | "name": "stdout", 344 | "output_type": "stream", 345 | "text": [ 346 | "branch 'main' set up to track 'origin/main'.\n", 347 | "\u001b[36;1m\u001b[1;3mCommitted to Github\u001b[0m\u001b[32;1m\u001b[1;3mThe commit with the message \"Testing LLM agent!\" has been successfully made to GitHub.\u001b[0m\n", 348 | "\n", 349 | "\u001b[1m> Finished chain.\u001b[0m\n" 350 | ] 351 | }, 352 | { 353 | "data": { 354 | "text/plain": [ 355 | "{'input': \"Create a github commit with the message 'Testing LLM agent!'\",\n", 356 | " 'output': 'The commit with the message \"Testing LLM agent!\" has been successfully made to GitHub.'}" 357 | ] 358 | }, 359 | "execution_count": 10, 360 | "metadata": {}, 361 | "output_type": "execute_result" 362 | } 363 | ], 364 | "source": [ 365 | "agent_executor.invoke({\"input\": \"Create a github commit with the message 'Testing LLM agent!'\"})" 366 | ] 367 | }, 368 | { 369 | "cell_type": "markdown", 370 | "metadata": {}, 371 | "source": [ 372 | "![](2024-08-26-12-39-39.png)" 373 | ] 374 | }, 375 | { 376 | "cell_type": "markdown", 377 | "metadata": {}, 378 | "source": [ 379 | "And voila, the agent calls the right function and creates the commit!\n", 380 | "\n", 381 | "(ignore the github related error, let's focus on the agent capabilities)." 382 | ] 383 | }, 384 | { 385 | "cell_type": "markdown", 386 | "metadata": {}, 387 | "source": [ 388 | "And voila! We made a commit!\n", 389 | "\n", 390 | "![](2023-10-30-13-04-20.png)" 391 | ] 392 | }, 393 | { 394 | "cell_type": "markdown", 395 | "metadata": {}, 396 | "source": [ 397 | "Ok, this is cool but, we can go beyond that, and add the ability to create a file and then commit that file to our desired repository. Let's do that!\n", 398 | "\n", 399 | "To do it all we have to do is amplify the Agent's toolkit with another tool that can create files in the current folder." 400 | ] 401 | }, 402 | { 403 | "cell_type": "markdown", 404 | "metadata": {}, 405 | "source": [ 406 | "Perfect! Now, let's add this to our github agent" 407 | ] 408 | }, 409 | { 410 | "cell_type": "code", 411 | "execution_count": 21, 412 | "metadata": {}, 413 | "outputs": [], 414 | "source": [ 415 | "@tool\n", 416 | "def create_file_tool(filename, contents):\n", 417 | " \"\"\"This function creates a file given its filename and its contents provided as inputs.\"\"\"\n", 418 | " with open(filename, \"w\") as f:\n", 419 | " f.write(contents)\n", 420 | " \n", 421 | " return \"File created\"" 422 | ] 423 | }, 424 | { 425 | "cell_type": "markdown", 426 | "metadata": {}, 427 | "source": [ 428 | "Now, let's add that functionality to our agent by just updating the tools list." 429 | ] 430 | }, 431 | { 432 | "cell_type": "code", 433 | "execution_count": 22, 434 | "metadata": {}, 435 | "outputs": [], 436 | "source": [ 437 | "tools = [github_commit_tool, create_file_tool]" 438 | ] 439 | }, 440 | { 441 | "cell_type": "code", 442 | "execution_count": 23, 443 | "metadata": {}, 444 | "outputs": [], 445 | "source": [ 446 | "agent = create_tool_calling_agent(llm, tools, prompt)" 447 | ] 448 | }, 449 | { 450 | "cell_type": "code", 451 | "execution_count": 24, 452 | "metadata": {}, 453 | "outputs": [], 454 | "source": [ 455 | "# Let's also inspect this agent execution in langsmith!\n", 456 | "agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)" 457 | ] 458 | }, 459 | { 460 | "cell_type": "markdown", 461 | "metadata": {}, 462 | "source": [ 463 | "Let's also inspect this agent execution in langsmith!" 464 | ] 465 | }, 466 | { 467 | "cell_type": "code", 468 | "execution_count": 26, 469 | "metadata": {}, 470 | "outputs": [ 471 | { 472 | "name": "stdout", 473 | "output_type": "stream", 474 | "text": [ 475 | "\n", 476 | "\n", 477 | "\u001b[1m> Entering new AgentExecutor chain...\u001b[0m\n", 478 | "\u001b[32;1m\u001b[1;3m\n", 479 | "Invoking: `create_file_tool` with `{'filename': 'pancakes-are-the-best-breakfast.md', 'contents': '# Why Pancakes are the Most Delicious Breakfast\\n\\n- **Fluffy Texture**: Pancakes have a light and fluffy texture that melts in your mouth.\\n- **Versatile Toppings**: They can be topped with a variety of delicious options like syrup, fruits, whipped cream, and nuts.\\n- **Quick to Make**: Pancakes are quick and easy to prepare, making them a perfect breakfast choice.\\n- **Comfort Food**: They provide a sense of comfort and nostalgia, often reminding us of family breakfasts.\\n- **Customizable**: You can easily customize the batter with flavors like chocolate, vanilla, or spices.\\n- **Great for Sharing**: Pancakes can be made in large batches, perfect for sharing with family and friends.\\n- **Satisfying**: They are filling and satisfying, providing a good start to the day.\\n- **Healthy Options**: You can make healthier versions using whole grains, fruits, or even protein powder.\\n- **Cultural Variations**: Pancakes come in many forms across different cultures, from crepes to flapjacks.\\n- **Fun to Make**: Cooking pancakes can be a fun activity, especially for kids who enjoy flipping them!'}`\n", 480 | "\n", 481 | "\n", 482 | "\u001b[0m\u001b[33;1m\u001b[1;3mFile created\u001b[0m\u001b[32;1m\u001b[1;3m\n", 483 | "Invoking: `github_commit_tool` with `{'commit_message': 'Add pancakes-are-the-best-breakfast.md with reasons why pancakes are the most delicious breakfast.'}`\n", 484 | "\n", 485 | "\n", 486 | "\u001b[0m[main abe2980] Add pancakes-are-the-best-breakfast.md with reasons why pancakes are the most delicious breakfast.\n", 487 | " 3 files changed, 30 insertions(+), 37 deletions(-)\n" 488 | ] 489 | }, 490 | { 491 | "name": "stderr", 492 | "output_type": "stream", 493 | "text": [ 494 | "To https://github.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain.git\n", 495 | " d5c44bc..abe2980 main -> main\n" 496 | ] 497 | }, 498 | { 499 | "name": "stdout", 500 | "output_type": "stream", 501 | "text": [ 502 | "branch 'main' set up to track 'origin/main'.\n", 503 | "\u001b[36;1m\u001b[1;3mCommitted to Github\u001b[0m\u001b[32;1m\u001b[1;3mI have created the file named \"pancakes-are-the-best-breakfast.md\" containing 10 bullet points on why pancakes are the most delicious breakfast, and it has been successfully committed to GitHub.\u001b[0m\n", 504 | "\n", 505 | "\u001b[1m> Finished chain.\u001b[0m\n" 506 | ] 507 | }, 508 | { 509 | "data": { 510 | "text/plain": [ 511 | "{'input': 'Create and commit a file named: \"pancakes-are-the-best-breakfast.md\" containing 10 bullet points on why pancakes are the most delicious breakfast.',\n", 512 | " 'output': 'I have created the file named \"pancakes-are-the-best-breakfast.md\" containing 10 bullet points on why pancakes are the most delicious breakfast, and it has been successfully committed to GitHub.'}" 513 | ] 514 | }, 515 | "execution_count": 26, 516 | "metadata": {}, 517 | "output_type": "execute_result" 518 | } 519 | ], 520 | "source": [ 521 | "agent_executor.invoke({'input': 'Create and commit a file named: \"pancakes-are-the-best-breakfast.md\" containing 10 bullet points on why\\\n", 522 | " pancakes are the most delicious breakfast.'})" 523 | ] 524 | }, 525 | { 526 | "cell_type": "markdown", 527 | "metadata": {}, 528 | "source": [ 529 | "We can inspect this more thoroughly in the [langsmith platform](https://smith.langchain.com/)." 530 | ] 531 | }, 532 | { 533 | "cell_type": "markdown", 534 | "metadata": {}, 535 | "source": [ 536 | "Ok, before we finish with this example, let's add a last tool to read files from a repo, for situations where we might write some Python code and we want our agent to integrate that Python code into our repository. " 537 | ] 538 | }, 539 | { 540 | "cell_type": "markdown", 541 | "metadata": {}, 542 | "source": [ 543 | "Again, we follow the same procedure:\n", 544 | "\n", 545 | "- Create the tool\n", 546 | "- Update our tookit\n", 547 | "- Update our agent\n", 548 | "- Run a test task to check if it works " 549 | ] 550 | }, 551 | { 552 | "cell_type": "markdown", 553 | "metadata": {}, 554 | "source": [ 555 | "I think this is a great usecase for agents like these even if sometimes the performance and reliability is a bit off, because now, its a matter of optimizing the edges to get it to work for complex use cases." 556 | ] 557 | }, 558 | { 559 | "cell_type": "markdown", 560 | "metadata": {}, 561 | "source": [ 562 | "# DISCLAIMER\n", 563 | "\n", 564 | "This example is for demonstration purposes, it does not put in place any measures to counter act potential security vulnerabilities\n", 565 | "associated with giving large language models the ability to execute shell commands in our behalf without authentication procedures\n", 566 | "and sandbox environments." 567 | ] 568 | } 569 | ], 570 | "metadata": { 571 | "kernelspec": { 572 | "display_name": "oreilly-langchain", 573 | "language": "python", 574 | "name": "oreilly-langchain" 575 | }, 576 | "language_info": { 577 | "codemirror_mode": { 578 | "name": "ipython", 579 | "version": 3 580 | }, 581 | "file_extension": ".py", 582 | "mimetype": "text/x-python", 583 | "name": "python", 584 | "nbconvert_exporter": "python", 585 | "pygments_lexer": "ipython3", 586 | "version": "3.11.9" 587 | } 588 | }, 589 | "nbformat": 4, 590 | "nbformat_minor": 2 591 | } 592 | -------------------------------------------------------------------------------- /notebooks/3.2-langchain-agent-with-langgraph.ipynb: -------------------------------------------------------------------------------- 1 | { 2 | "cells": [ 3 | { 4 | "cell_type": "code", 5 | "execution_count": null, 6 | "metadata": {}, 7 | "outputs": [], 8 | "source": [] 9 | }, 10 | { 11 | "cell_type": "code", 12 | "execution_count": 1, 13 | "metadata": {}, 14 | "outputs": [], 15 | "source": [ 16 | "# Import relevant functionality\n", 17 | "from langchain_openai import ChatOpenAI\n", 18 | "from langchain_community.tools.tavily_search import TavilySearchResults\n", 19 | "from langchain_core.messages import HumanMessage\n", 20 | "from langgraph.checkpoint.memory import MemorySaver\n", 21 | "from langgraph.prebuilt import create_react_agent" 22 | ] 23 | }, 24 | { 25 | "cell_type": "code", 26 | "execution_count": 2, 27 | "metadata": {}, 28 | "outputs": [], 29 | "source": [ 30 | "# Create the agent\n", 31 | "memory = MemorySaver()" 32 | ] 33 | }, 34 | { 35 | "cell_type": "code", 36 | "execution_count": 3, 37 | "metadata": {}, 38 | "outputs": [ 39 | { 40 | "data": { 41 | "text/plain": [ 42 | "AIMessage(content='An agent based on large language models (LLMs) refers to a software entity or system that utilizes the capabilities of LLMs to perform tasks, interact with users, and make decisions based on natural language input. These agents can be designed for a variety of applications, including customer support, content generation, virtual assistants, and more. Here are some key characteristics and functionalities of such agents:\\n\\n1. **Natural Language Understanding (NLU)**: Agents leverage LLMs to comprehend and interpret user inputs in natural language, allowing for more intuitive interactions.\\n\\n2. **Natural Language Generation (NLG)**: They can generate human-like text responses, making conversations more engaging and contextually relevant.\\n\\n3. **Context Awareness**: Advanced agents can maintain context over a conversation, enabling more coherent and meaningful interactions over multiple exchanges.\\n\\n4. **Task Execution**: Agents may perform specific tasks such as answering questions, providing recommendations, or even executing commands, depending on their design and the context of the interaction.\\n\\n5. **Learning and Adaptation**: Some agents can learn from user interactions over time, improving their responses and functionalities based on feedback and usage patterns.\\n\\n6. **Integration with Other Systems**: Agents can often be integrated with other software systems and databases to access real-time information, manage tasks, or pull in relevant data to enhance their responses.\\n\\n7. **Personalization**: They can be designed to tailor responses based on user preferences or previous interactions, creating a more personalized experience.\\n\\nExamples of agents based on large language models include chatbots, virtual personal assistants like Siri or Google Assistant, and automated customer service representatives. These agents are increasingly being used across various industries to enhance user experience and streamline operations.', additional_kwargs={'refusal': None}, response_metadata={'token_usage': {'completion_tokens': 343, 'prompt_tokens': 17, 'total_tokens': 360, 'completion_tokens_details': {'audio_tokens': 0, 'reasoning_tokens': 0, 'accepted_prediction_tokens': 0, 'rejected_prediction_tokens': 0}, 'prompt_tokens_details': {'audio_tokens': 0, 'cached_tokens': 0}}, 'model_name': 'gpt-4o-mini-2024-07-18', 'system_fingerprint': 'fp_06737a9306', 'finish_reason': 'stop', 'logprobs': None}, id='run-cf74da99-bcea-44ae-a8ce-dcc26ac0d85b-0', usage_metadata={'input_tokens': 17, 'output_tokens': 343, 'total_tokens': 360, 'input_token_details': {'audio': 0, 'cache_read': 0}, 'output_token_details': {'audio': 0, 'reasoning': 0}})" 43 | ] 44 | }, 45 | "execution_count": 3, 46 | "metadata": {}, 47 | "output_type": "execute_result" 48 | } 49 | ], 50 | "source": [ 51 | "llm = ChatOpenAI(model_name=\"gpt-4o-mini\")\n", 52 | "\n", 53 | "\n", 54 | "llm.invoke(\"What is an agent based on large language models?\")" 55 | ] 56 | }, 57 | { 58 | "cell_type": "code", 59 | "execution_count": 4, 60 | "metadata": {}, 61 | "outputs": [], 62 | "source": [ 63 | "search = TavilySearchResults(max_results=3)" 64 | ] 65 | }, 66 | { 67 | "cell_type": "code", 68 | "execution_count": 5, 69 | "metadata": {}, 70 | "outputs": [], 71 | "source": [ 72 | "tools = [search]" 73 | ] 74 | }, 75 | { 76 | "cell_type": "code", 77 | "execution_count": 6, 78 | "metadata": {}, 79 | "outputs": [ 80 | { 81 | "ename": "NameError", 82 | "evalue": "name 'agent_executor' is not defined", 83 | "output_type": "error", 84 | "traceback": [ 85 | "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m", 86 | "\u001b[0;31mNameError\u001b[0m Traceback (most recent call last)", 87 | "Cell \u001b[0;32mIn[6], line 1\u001b[0m\n\u001b[0;32m----> 1\u001b[0m \u001b[43magent_executor\u001b[49m\n", 88 | "\u001b[0;31mNameError\u001b[0m: name 'agent_executor' is not defined" 89 | ] 90 | } 91 | ], 92 | "source": [ 93 | "agent_executor" 94 | ] 95 | }, 96 | { 97 | "cell_type": "code", 98 | "execution_count": 7, 99 | "metadata": {}, 100 | "outputs": [], 101 | "source": [ 102 | "agent_executor = create_react_agent(llm, tools, checkpointer=memory)" 103 | ] 104 | }, 105 | { 106 | "cell_type": "code", 107 | "execution_count": 8, 108 | "metadata": {}, 109 | "outputs": [ 110 | { 111 | "name": "stdout", 112 | "output_type": "stream", 113 | "text": [ 114 | "{'agent': {'messages': [AIMessage(content=\"Hi Bob! It's great to meet you. How's life in San Francisco?\", additional_kwargs={'refusal': None}, response_metadata={'token_usage': {'completion_tokens': 17, 'prompt_tokens': 88, 'total_tokens': 105, 'completion_tokens_details': {'audio_tokens': 0, 'reasoning_tokens': 0, 'accepted_prediction_tokens': 0, 'rejected_prediction_tokens': 0}, 'prompt_tokens_details': {'audio_tokens': 0, 'cached_tokens': 0}}, 'model_name': 'gpt-4o-mini-2024-07-18', 'system_fingerprint': 'fp_bba3c8e70b', 'finish_reason': 'stop', 'logprobs': None}, id='run-af9de20a-f905-4972-bfc5-5ec5b707ff4b-0', usage_metadata={'input_tokens': 88, 'output_tokens': 17, 'total_tokens': 105, 'input_token_details': {'audio': 0, 'cache_read': 0}, 'output_token_details': {'audio': 0, 'reasoning': 0}})]}}\n", 115 | "----\n", 116 | "{'agent': {'messages': [AIMessage(content='', additional_kwargs={'tool_calls': [{'id': 'call_UFdwE6U5TjsJpVKI40kHzmHu', 'function': {'arguments': '{\"query\":\"San Francisco weather\"}', 'name': 'tavily_search_results_json'}, 'type': 'function'}], 'refusal': None}, response_metadata={'token_usage': {'completion_tokens': 20, 'prompt_tokens': 120, 'total_tokens': 140, 'completion_tokens_details': {'audio_tokens': 0, 'reasoning_tokens': 0, 'accepted_prediction_tokens': 0, 'rejected_prediction_tokens': 0}, 'prompt_tokens_details': {'audio_tokens': 0, 'cached_tokens': 0}}, 'model_name': 'gpt-4o-mini-2024-07-18', 'system_fingerprint': 'fp_bba3c8e70b', 'finish_reason': 'tool_calls', 'logprobs': None}, id='run-809ac445-45e1-45b7-9aec-33bbcf059ec9-0', tool_calls=[{'name': 'tavily_search_results_json', 'args': {'query': 'San Francisco weather'}, 'id': 'call_UFdwE6U5TjsJpVKI40kHzmHu', 'type': 'tool_call'}], usage_metadata={'input_tokens': 120, 'output_tokens': 20, 'total_tokens': 140, 'input_token_details': {'audio': 0, 'cache_read': 0}, 'output_token_details': {'audio': 0, 'reasoning': 0}})]}}\n", 117 | "----\n", 118 | "{'tools': {'messages': [ToolMessage(content='[{\"url\": \"https://www.weatherapi.com/\", \"content\": \"{\\'location\\': {\\'name\\': \\'San Francisco\\', \\'region\\': \\'California\\', \\'country\\': \\'United States of America\\', \\'lat\\': 37.775, \\'lon\\': -122.4183, \\'tz_id\\': \\'America/Los_Angeles\\', \\'localtime_epoch\\': 1733951474, \\'localtime\\': \\'2024-12-11 13:11\\'}, \\'current\\': {\\'last_updated_epoch\\': 1733950800, \\'last_updated\\': \\'2024-12-11 13:00\\', \\'temp_c\\': 12.2, \\'temp_f\\': 54.0, \\'is_day\\': 1, \\'condition\\': {\\'text\\': \\'Partly cloudy\\', \\'icon\\': \\'//cdn.weatherapi.com/weather/64x64/day/116.png\\', \\'code\\': 1003}, \\'wind_mph\\': 4.3, \\'wind_kph\\': 6.8, \\'wind_degree\\': 43, \\'wind_dir\\': \\'NE\\', \\'pressure_mb\\': 1021.0, \\'pressure_in\\': 30.15, \\'precip_mm\\': 0.0, \\'precip_in\\': 0.0, \\'humidity\\': 66, \\'cloud\\': 75, \\'feelslike_c\\': 11.8, \\'feelslike_f\\': 53.3, \\'windchill_c\\': 7.7, \\'windchill_f\\': 45.8, \\'heatindex_c\\': 8.8, \\'heatindex_f\\': 47.9, \\'dewpoint_c\\': 6.7, \\'dewpoint_f\\': 44.0, \\'vis_km\\': 16.0, \\'vis_miles\\': 9.0, \\'uv\\': 1.8, \\'gust_mph\\': 5.7, \\'gust_kph\\': 9.2}}\"}, {\"url\": \"https://www.msn.com/en-us/weather/topstories/december-11-2024-san-francisco-bay-area-weather-forecast/vi-AA1vFB5g\", \"content\": \"KRON San Francisco. December 11, 2024 San Francisco Bay Area weather forecast. Posted: December 11, 2024 | Last updated: December 11, 2024. KRON4 Meteorologist John Shrable has the latest updates\"}, {\"url\": \"https://weatherspark.com/h/m/557/2024/11/Historical-Weather-in-November-2024-in-San-Francisco-California-United-States\", \"content\": \"This report shows the past weather for San Francisco, providing a weather history for November 2024. It features all historical weather data series we have available, including the San Francisco temperature history for November 2024. San Francisco Temperature History November 2024 Hourly Temperature in November 2024 in San Francisco Cloud Cover in November 2024 in San Francisco Daily Precipitation in November 2024 in San Francisco Observed Weather in November 2024 in San Francisco Hours of Daylight and Twilight in November 2024 in San Francisco Humidity Comfort Levels in November 2024 in San Francisco Wind Speed in November 2024 in San Francisco Hourly Wind Speed in November 2024 in San Francisco The details of the data sources used for this report can be found on the San Francisco International Airport page.\"}]', name='tavily_search_results_json', id='22ffeb9f-caf6-4d30-ba73-19add990aa4f', tool_call_id='call_UFdwE6U5TjsJpVKI40kHzmHu', artifact={'query': 'San Francisco weather', 'follow_up_questions': None, 'answer': None, 'images': [], 'results': [{'title': 'Weather in San Francisco', 'url': 'https://www.weatherapi.com/', 'content': \"{'location': {'name': 'San Francisco', 'region': 'California', 'country': 'United States of America', 'lat': 37.775, 'lon': -122.4183, 'tz_id': 'America/Los_Angeles', 'localtime_epoch': 1733951474, 'localtime': '2024-12-11 13:11'}, 'current': {'last_updated_epoch': 1733950800, 'last_updated': '2024-12-11 13:00', 'temp_c': 12.2, 'temp_f': 54.0, 'is_day': 1, 'condition': {'text': 'Partly cloudy', 'icon': '//cdn.weatherapi.com/weather/64x64/day/116.png', 'code': 1003}, 'wind_mph': 4.3, 'wind_kph': 6.8, 'wind_degree': 43, 'wind_dir': 'NE', 'pressure_mb': 1021.0, 'pressure_in': 30.15, 'precip_mm': 0.0, 'precip_in': 0.0, 'humidity': 66, 'cloud': 75, 'feelslike_c': 11.8, 'feelslike_f': 53.3, 'windchill_c': 7.7, 'windchill_f': 45.8, 'heatindex_c': 8.8, 'heatindex_f': 47.9, 'dewpoint_c': 6.7, 'dewpoint_f': 44.0, 'vis_km': 16.0, 'vis_miles': 9.0, 'uv': 1.8, 'gust_mph': 5.7, 'gust_kph': 9.2}}\", 'score': 0.9992536, 'raw_content': None}, {'title': 'December 11, 2024 San Francisco Bay Area weather forecast - MSN', 'url': 'https://www.msn.com/en-us/weather/topstories/december-11-2024-san-francisco-bay-area-weather-forecast/vi-AA1vFB5g', 'content': 'KRON San Francisco. December 11, 2024 San Francisco Bay Area weather forecast. Posted: December 11, 2024 | Last updated: December 11, 2024. KRON4 Meteorologist John Shrable has the latest updates', 'score': 0.9957684, 'raw_content': None}, {'title': 'November 2024 Weather History in San Francisco', 'url': 'https://weatherspark.com/h/m/557/2024/11/Historical-Weather-in-November-2024-in-San-Francisco-California-United-States', 'content': 'This report shows the past weather for San Francisco, providing a weather history for November 2024. It features all historical weather data series we have available, including the San Francisco temperature history for November 2024. San Francisco Temperature History November 2024 Hourly Temperature in November 2024 in San Francisco Cloud Cover in November 2024 in San Francisco Daily Precipitation in November 2024 in San Francisco Observed Weather in November 2024 in San Francisco Hours of Daylight and Twilight in November 2024 in San Francisco Humidity Comfort Levels in November 2024 in San Francisco Wind Speed in November 2024 in San Francisco Hourly Wind Speed in November 2024 in San Francisco The details of the data sources used for this report can be found on the San Francisco International Airport page.', 'score': 0.99021614, 'raw_content': None}], 'response_time': 2.74})]}}\n", 119 | "----\n", 120 | "{'agent': {'messages': [AIMessage(content='The current weather in San Francisco is partly cloudy with a temperature of about 54°F (12.2°C). Here are some additional details:\\n\\n- **Humidity**: 66%\\n- **Wind**: Northeastern at 4.3 mph (6.8 kph)\\n- **Visibility**: 16 km (9 miles)\\n- **Pressure**: 30.15 inHg (1021.0 mb)\\n\\nIt feels slightly cooler at around 53.3°F (11.8°C) due to the wind chill.\\n\\nIf you want more detailed or updated information, you can check out [WeatherAPI](https://www.weatherapi.com/) or local news outlets.', additional_kwargs={'refusal': None}, response_metadata={'token_usage': {'completion_tokens': 140, 'prompt_tokens': 867, 'total_tokens': 1007, 'completion_tokens_details': {'audio_tokens': 0, 'reasoning_tokens': 0, 'accepted_prediction_tokens': 0, 'rejected_prediction_tokens': 0}, 'prompt_tokens_details': {'audio_tokens': 0, 'cached_tokens': 0}}, 'model_name': 'gpt-4o-mini-2024-07-18', 'system_fingerprint': 'fp_bba3c8e70b', 'finish_reason': 'stop', 'logprobs': None}, id='run-7c5358f7-2be9-4753-8375-1bb0470c58c5-0', usage_metadata={'input_tokens': 867, 'output_tokens': 140, 'total_tokens': 1007, 'input_token_details': {'audio': 0, 'cache_read': 0}, 'output_token_details': {'audio': 0, 'reasoning': 0}})]}}\n", 121 | "----\n" 122 | ] 123 | } 124 | ], 125 | "source": [ 126 | "# Use the agent\n", 127 | "config = {\"configurable\": {\"thread_id\": \"abc123\"}}\n", 128 | "for chunk in agent_executor.stream(\n", 129 | " {\"messages\": [HumanMessage(content=\"hi im bob! and i live in sf\")]}, config\n", 130 | "):\n", 131 | " print(chunk)\n", 132 | " print(\"----\")\n", 133 | "\n", 134 | "for chunk in agent_executor.stream(\n", 135 | " {\"messages\": [HumanMessage(content=\"whats the weather where I live?\")]}, config\n", 136 | "):\n", 137 | " print(chunk)\n", 138 | " print(\"----\")" 139 | ] 140 | }, 141 | { 142 | "cell_type": "code", 143 | "execution_count": null, 144 | "metadata": {}, 145 | "outputs": [], 146 | "source": [] 147 | } 148 | ], 149 | "metadata": { 150 | "kernelspec": { 151 | "display_name": "oreilly-langchain", 152 | "language": "python", 153 | "name": "oreilly-langchain" 154 | }, 155 | "language_info": { 156 | "codemirror_mode": { 157 | "name": "ipython", 158 | "version": 3 159 | }, 160 | "file_extension": ".py", 161 | "mimetype": "text/x-python", 162 | "name": "python", 163 | "nbconvert_exporter": "python", 164 | "pygments_lexer": "ipython3", 165 | "version": "3.11.9" 166 | } 167 | }, 168 | "nbformat": 4, 169 | "nbformat_minor": 2 170 | } 171 | -------------------------------------------------------------------------------- /notebooks/6.0-live-demo-chat-with-langchain-docs-urls.ipynb: -------------------------------------------------------------------------------- 1 | { 2 | "cells": [ 3 | { 4 | "cell_type": "code", 5 | "execution_count": null, 6 | "metadata": {}, 7 | "outputs": [], 8 | "source": [ 9 | "urls = ['https://python.langchain.com/docs/concepts/#__docusaurus_skipToContent_fallback', \n", 10 | " 'https://python.langchain.com/', 'https://python.langchain.com/docs/integrations/providers/', \n", 11 | " 'https://python.langchain.com/api_reference/', 'https://python.langchain.com/docs/concepts/#', \n", 12 | " 'https://python.langchain.com/docs/contributing/', 'https://python.langchain.com/docs/people/', \n", 13 | " 'https://python.langchain.com/docs/troubleshooting/errors/', 'https://docs.smith.langchain.com/', \n", 14 | " 'https://langchain-ai.github.io/langgraph/', 'https://smith.langchain.com/hub', 'https://js.langchain.com/',\n", 15 | " 'https://python.langchain.com/docs/introduction/', 'https://python.langchain.com/v0.2/docs/introduction',\n", 16 | " 'https://python.langchain.com/v0.1/docs/get_started/introduction', 'https://chat.langchain.com/', \n", 17 | " 'https://github.com/langchain-ai/langchain', 'https://python.langchain.com/docs/tutorials/', \n", 18 | " 'https://python.langchain.com/docs/tutorials/graph/', 'https://python.langchain.com/docs/tutorials/llm_chain/',\n", 19 | " 'https://python.langchain.com/docs/tutorials/chatbot/', 'https://python.langchain.com/docs/tutorials/qa_chat_history/', \n", 20 | " 'https://python.langchain.com/docs/tutorials/extraction/', 'https://python.langchain.com/docs/tutorials/agents/', \n", 21 | " 'https://python.langchain.com/docs/tutorials/classification/', 'https://python.langchain.com/docs/tutorials/rag/',\n", 22 | " 'https://python.langchain.com/docs/tutorials/retrievers/', 'https://python.langchain.com/docs/tutorials/sql_qa/', \n", 23 | " 'https://python.langchain.com/docs/tutorials/summarization/', 'https://python.langchain.com/docs/how_to/', \n", 24 | " 'https://python.langchain.com/docs/how_to/tools_chain/', 'https://python.langchain.com/docs/how_to/vectorstore_retriever/',\n", 25 | " 'https://python.langchain.com/docs/how_to/chatbots_memory/', 'https://python.langchain.com/docs/how_to/example_selectors/', \n", 26 | " 'https://python.langchain.com/docs/how_to/graph_semantic/', 'https://python.langchain.com/docs/how_to/parallel/', \n", 27 | " 'https://python.langchain.com/docs/how_to/chat_streaming/', 'https://python.langchain.com/docs/how_to/binding/', \n", 28 | " 'https://python.langchain.com/docs/how_to/chatbots_retrieval/', 'https://python.langchain.com/docs/how_to/few_shot_examples_chat/', \n", 29 | " 'https://python.langchain.com/docs/how_to/function_calling/', 'https://python.langchain.com/docs/how_to/installation/', \n", 30 | " 'https://python.langchain.com/docs/how_to/query_few_shot/', 'https://python.langchain.com/docs/how_to/few_shot_examples/', \n", 31 | " 'https://python.langchain.com/docs/how_to/functions/', 'https://python.langchain.com/docs/how_to/output_parser_structured/', \n", 32 | " 'https://python.langchain.com/docs/how_to/query_no_queries/', 'https://python.langchain.com/docs/how_to/routing/', 'https://python.langchain.com/docs/how_to/structured_output/', 'https://python.langchain.com/docs/how_to/summarize_map_reduce/', 'https://python.langchain.com/docs/how_to/summarize_refine/', 'https://python.langchain.com/docs/how_to/summarize_stuff/', 'https://python.langchain.com/docs/how_to/toolkits/', 'https://python.langchain.com/docs/how_to/tools_prompting/', 'https://python.langchain.com/docs/how_to/agent_executor/', 'https://python.langchain.com/docs/how_to/graph_constructing/', 'https://python.langchain.com/docs/how_to/prompts_partial/', 'https://python.langchain.com/docs/how_to/query_multiple_queries/', 'https://python.langchain.com/docs/how_to/tools_builtin/', 'https://python.langchain.com/docs/how_to/passthrough/',\n", 33 | " 'https://python.langchain.com/docs/how_to/prompts_composition/', 'https://python.langchain.com/docs/how_to/query_multiple_retrievers/', 'https://python.langchain.com/docs/how_to/assign/', 'https://python.langchain.com/docs/how_to/query_constructing_filters/', 'https://python.langchain.com/docs/how_to/configure/', 'https://python.langchain.com/docs/how_to/query_high_cardinality/', 'https://python.langchain.com/docs/how_to/document_loader_custom/', 'https://python.langchain.com/docs/how_to/HTML_header_metadata_splitter/', 'https://python.langchain.com/docs/how_to/HTML_section_aware_splitter/', 'https://python.langchain.com/docs/how_to/MultiQueryRetriever/', 'https://python.langchain.com/docs/how_to/add_scores_retriever/', 'https://python.langchain.com/docs/how_to/caching_embeddings/', 'https://python.langchain.com/docs/how_to/callbacks_async/', 'https://python.langchain.com/docs/how_to/callbacks_attach/', 'https://python.langchain.com/docs/how_to/callbacks_constructor/', 'https://python.langchain.com/docs/how_to/callbacks_custom_events/', 'https://python.langchain.com/docs/how_to/callbacks_runtime/', 'https://python.langchain.com/docs/how_to/character_text_splitter/', 'https://python.langchain.com/docs/how_to/chat_model_caching/', 'https://python.langchain.com/docs/how_to/chat_model_rate_limiting/', 'https://python.langchain.com/docs/how_to/chat_models_universal_init/', 'https://python.langchain.com/docs/how_to/chat_token_usage_tracking/', 'https://python.langchain.com/docs/how_to/chatbots_tools/', 'https://python.langchain.com/docs/how_to/code_splitter/', 'https://python.langchain.com/docs/how_to/contextual_compression/', 'https://python.langchain.com/docs/how_to/convert_runnable_to_tool/', 'https://python.langchain.com/docs/how_to/custom_callbacks/',\n", 34 | " 'https://python.langchain.com/docs/how_to/custom_chat_model/', 'https://python.langchain.com/docs/how_to/custom_llm/', 'https://python.langchain.com/docs/how_to/custom_retriever/', 'https://python.langchain.com/docs/how_to/custom_tools/', 'https://python.langchain.com/docs/how_to/debugging/', 'https://python.langchain.com/docs/how_to/document_loader_csv/', 'https://python.langchain.com/docs/how_to/document_loader_directory/', 'https://python.langchain.com/docs/how_to/document_loader_html/', 'https://python.langchain.com/docs/how_to/document_loader_json/', 'https://python.langchain.com/docs/how_to/document_loader_markdown/', 'https://python.langchain.com/docs/how_to/document_loader_office_file/', 'https://python.langchain.com/docs/how_to/document_loader_pdf/']" 35 | ] 36 | } 37 | ], 38 | "metadata": { 39 | "language_info": { 40 | "name": "python" 41 | } 42 | }, 43 | "nbformat": 4, 44 | "nbformat_minor": 2 45 | } 46 | -------------------------------------------------------------------------------- /notebooks/README.md: -------------------------------------------------------------------------------- 1 | # O'Reilly Live Trainining - Getting Started with Langchain 2 | 3 | Link to the event: https://learning.oreilly.com/live-events/getting-started-with-langchain/0636920098586/ 4 | 5 | ## Setup 6 | 7 | **Conda** 8 | 9 | - Install [anaconda](https://www.anaconda.com/download) 10 | - Create an environment: `conda create -n oreilly-langchain` 11 | - Activate your environment with: `conda activate oreilly-langchain` 12 | - Install requirements with: 13 | - `pip install -r ./notebooks/requirements.txt` 14 | 15 | - Setup your openai [API key](https://platform.openai.com/) 16 | 17 | **Pip** 18 | 19 | 20 | 1. **Create a Virtual Environment:** 21 | Navigate to your project directory. If using Python 3's built-in `venv`: 22 | ```bash 23 | python -m venv oreilly_env 24 | ``` 25 | If you're using `virtualenv`: 26 | ```bash 27 | virtualenv oreilly_env 28 | ``` 29 | 30 | 2. **Activate the Virtual Environment:** 31 | - **On Windows:** 32 | ```bash 33 | .\oreilly_env\Scripts\activate 34 | ``` 35 | - **On macOS and Linux:** 36 | ```bash 37 | source oreilly_env/bin/activate 38 | ``` 39 | 40 | 3. **Install Dependencies from `requirements.txt`:** 41 | ```bash 42 | pip install python-dotenv 43 | pip install -r requirements.txt 44 | ``` 45 | 46 | 4. Setup your openai [API key](https://platform.openai.com/) 47 | 48 | Remember to deactivate the virtual environment once you're done by simply typing: 49 | ```bash 50 | deactivate 51 | ``` 52 | 53 | ## Setup your .env file 54 | 55 | - Change the `.env.example` file to `.env` and add your OpenAI API key. 56 | 57 | ## To use this Environment with Jupyter Notebooks: 58 | 59 | ```python3 -m ipykernel install --user --name=oreilly-langchain``` -------------------------------------------------------------------------------- /notebooks/assets-resources/Augmenting-Human-Intellect_ A-Conceptual-Framework-1962-Doug Engelbart.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/Augmenting-Human-Intellect_ A-Conceptual-Framework-1962-Doug Engelbart.pdf -------------------------------------------------------------------------------- /notebooks/assets-resources/agent-loop-langgraph-version.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/agent-loop-langgraph-version.png -------------------------------------------------------------------------------- /notebooks/assets-resources/agent-loop.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/agent-loop.png -------------------------------------------------------------------------------- /notebooks/assets-resources/agent_loop.svg: -------------------------------------------------------------------------------- 1 |
User
User
Action/Tools
Action/Tools
LLM
LLM
Observation
Observation
Output
OutputText is not SVG - cannot display -------------------------------------------------------------------------------- /notebooks/assets-resources/ai.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/ai.png -------------------------------------------------------------------------------- /notebooks/assets-resources/attention-paper.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/attention-paper.pdf -------------------------------------------------------------------------------- /notebooks/assets-resources/attributes_runnable.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/attributes_runnable.png -------------------------------------------------------------------------------- /notebooks/assets-resources/basic-agent-diagram.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/basic-agent-diagram.png -------------------------------------------------------------------------------- /notebooks/assets-resources/chatgpt-demo.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/chatgpt-demo.png -------------------------------------------------------------------------------- /notebooks/assets-resources/components_input_output.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/components_input_output.png -------------------------------------------------------------------------------- /notebooks/assets-resources/components_input_type_output_type.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/components_input_type_output_type.png -------------------------------------------------------------------------------- /notebooks/assets-resources/diagram_langchain_agent_find_source_code.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/diagram_langchain_agent_find_source_code.png -------------------------------------------------------------------------------- /notebooks/assets-resources/embeddings-similarity.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/embeddings-similarity.png -------------------------------------------------------------------------------- /notebooks/assets-resources/embeddings-similarity2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/embeddings-similarity2.png -------------------------------------------------------------------------------- /notebooks/assets-resources/embeddings.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/embeddings.png -------------------------------------------------------------------------------- /notebooks/assets-resources/google-icon.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/google-icon.png -------------------------------------------------------------------------------- /notebooks/assets-resources/google-translate.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/google-translate.png -------------------------------------------------------------------------------- /notebooks/assets-resources/images_tbn.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/images_tbn.jpg -------------------------------------------------------------------------------- /notebooks/assets-resources/langchain-project-structure.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/langchain-project-structure.png -------------------------------------------------------------------------------- /notebooks/assets-resources/langchain-toolkits.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/langchain-toolkits.png -------------------------------------------------------------------------------- /notebooks/assets-resources/langgraph-components.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/langgraph-components.png -------------------------------------------------------------------------------- /notebooks/assets-resources/lcel-image.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/lcel-image.png -------------------------------------------------------------------------------- /notebooks/assets-resources/llm_paper_know_dont_know.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/llm_paper_know_dont_know.pdf -------------------------------------------------------------------------------- /notebooks/assets-resources/llm_predicts_pancakes.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/llm_predicts_pancakes.png -------------------------------------------------------------------------------- /notebooks/assets-resources/pancake_maker.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/pancake_maker.png -------------------------------------------------------------------------------- /notebooks/assets-resources/paper-llm-components.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/paper-llm-components.pdf -------------------------------------------------------------------------------- /notebooks/assets-resources/pydantic.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/pydantic.png -------------------------------------------------------------------------------- /notebooks/assets-resources/rag-docs.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/rag-docs.png -------------------------------------------------------------------------------- /notebooks/assets-resources/rag-langchain-retrieval.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/rag-langchain-retrieval.png -------------------------------------------------------------------------------- /notebooks/assets-resources/rag-langchain.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/rag-langchain.png -------------------------------------------------------------------------------- /notebooks/assets-resources/superheroes.csv: -------------------------------------------------------------------------------- 1 | Superhero Name,Superpower,Power Level,Catchphrase 2 | Captain Thunder,Bolt Manipulation,90,"Feel the power of the storm!" 3 | Silver Falcon,Flight and Agility,85,"Soar high, fearlessly!" 4 | Mystic Shadow,Invisibility and Illusions,78,"Disappear into the darkness!" 5 | Blaze Runner,Pyrokinesis,88,"Burn bright and fierce!" 6 | Electra-Wave,Electric Manipulation,82,"Unleash the electric waves!" 7 | Crimson Cyclone,Super Speed,91,"Blazing fast and unstoppable!" 8 | Aqua Fury,Hydrokinesis,80,"Ride the waves of power!" 9 | Lunar Guardian,Lunar Manipulation,77,"Embrace the moon's might!" 10 | Steel Titan,Super Strength and Durability,95,"Indestructible force of nature!" 11 | Nightblade,Night Vision and Stealth,84,"Strike from the shadows!" 12 | Frostbite,Ice Manipulation,87,"Chill your bones!" 13 | Starburst,Energy Projection,83,"Ignite the cosmos!" 14 | Sonic Dash,Sound Manipulation,86,"Hear the sound of victory!" 15 | Nova Surge,Energy Absorption and Redirection,89,"Harness the energy within!" 16 | Shadowcat,Intangibility,76,"Phase through the impossible!" 17 | Neon Spark,Light Manipulation,79,"Glow and dazzle!" 18 | Phoenix Flame,Resurrection and Healing,94,"Rise from the ashes!" 19 | Crystal Guardian,Crystallokinesis,81,"Shatter your illusions!" 20 | Earthshaker,Geokinesis,92,"Feel the earth tremble beneath you!" 21 | Silver Seraph,Telekinesis,85,"Move objects with your mind!" 22 | Stormbringer,Weather Manipulation,93,"Unleash the storm's fury!" 23 | Scarlet Siren,Mind Control,88,"Obey my commands!" 24 | Rift Walker,Dimensional Travel,90,"Step between worlds!" 25 | Chrono-Wizard,Time Manipulation,91,"Master of time's flow!" 26 | Blazing Comet,Fireball Projection,82,"Burn brighter than a comet!" 27 | Phantom Wisp,Invisibility,75,"Disappear like a wisp in the wind!" 28 | Luminous Knight,Luminokinesis,78,"Illuminate the darkest night!" 29 | Gravity Shift,Gravity Manipulation,89,"Bend gravity to your will!" 30 | Solar Empress,Solar Energy Absorption,87,"Harness the power of the sun!" 31 | Twilight Specter,Shadow Manipulation,80,"Dance with shadows!" 32 | Thunderstrike,Lightning Control,91,"Electrify the battlefield!" 33 | Nebula Weaver,Reality Warping,96,"Shape the fabric of reality!" 34 | Frostglide,Cryo-Teleportation,85,"Freeze and glide through space!" 35 | Zenith Blaze,Heat Vision,83,"Gaze into the flames of justice!" 36 | Astral Drifter,Astral Projection,79,"Roam the astral plane!" 37 | Blade Dancer,Swordsmanship and Agility,88,"Dance with deadly grace!" 38 | Azure Tempest,Water Manipulation,82,"Unleash the tempest's rage!" 39 | Ghost Sentinel,Intangibility and Invisibility,76,"Haunt your nightmares!" 40 | Ember Fox,Pyrokinetic Fox Shapeshifter,89,"Outfox your enemies with fire!" 41 | Psy-Bender,Telepathy,84,"Read minds like an open book!" 42 | Celestial Sphinx,Cosmic Awareness,93,"Unravel the secrets of the universe!" 43 | Dragonfist,Dragon Summoning and Martial Arts,92,"Unleash the dragon's fury!" 44 | Solar Flare,Solar Energy Projection,85,"Feel the burning light!" 45 | Night Lotus,Darkness Manipulation,78,"Bloom in the shadows!" 46 | Quantum Strider,Quantum Manipulation,90,"Walk the edge of reality!" 47 | Ironclad,Invulnerability and Enhanced Strength,95,"Invincible and mighty!" 48 | Shadow Stalker,Shadow Shifting,81,"Disappear and reappear at will!" 49 | Aqua Archer,Water Arrow Projection,80,"Shoot through water's flow!" 50 | Crystal Gazer,Crystal Ball Scrying,77,"See what the future holds!" 51 | -------------------------------------------------------------------------------- /notebooks/assets-resources/tagging.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/tagging.png -------------------------------------------------------------------------------- /notebooks/assets-resources/vectordb.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/assets-resources/vectordb.png -------------------------------------------------------------------------------- /notebooks/dev-notebooks/2023-12-13-14-12-45.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/dev-notebooks/2023-12-13-14-12-45.png -------------------------------------------------------------------------------- /notebooks/dev-notebooks/2023-12-13-14-25-37.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/dev-notebooks/2023-12-13-14-25-37.png -------------------------------------------------------------------------------- /notebooks/dev-notebooks/5.1-demo-playwright-scrape-articles.py: -------------------------------------------------------------------------------- 1 | # requires installing playwright! 2 | import pprint 3 | from langchain.text_splitter import RecursiveCharacterTextSplitter 4 | from langchain.document_loaders import AsyncChromiumLoader 5 | from langchain.document_transformers import BeautifulSoupTransformer 6 | from langchain.chains import create_extraction_chain 7 | from langchain_openai import ChatOpenAI 8 | 9 | def extract(content: str, schema: dict): 10 | return create_extraction_chain(schema=schema, llm=llm).run(content) 11 | 12 | def scrape_with_playwright(urls, schema): 13 | loader = AsyncChromiumLoader(urls) 14 | docs = loader.load() 15 | bs_transformer = BeautifulSoupTransformer() 16 | docs_transformed = bs_transformer.transform_documents(docs, tags_to_extract=["span"]) 17 | print("Extracting content with LLM") 18 | 19 | # Grab the first 1000 tokens of the site 20 | splitter = RecursiveCharacterTextSplitter.from_tiktoken_encoder(chunk_size=1000, chunk_overlap=0) 21 | splits = splitter.split_documents(docs_transformed) 22 | 23 | # Process the first split 24 | extracted_content = extract(schema=schema, content=splits[0].page_content) 25 | pprint.pprint(extracted_content) 26 | return extracted_content 27 | 28 | schema = { 29 | "properties": { 30 | "news_article_title": {"type": "string"}, 31 | "news_article_summary": {"type": "string"}, 32 | }, 33 | "required": ["news_article_title", "news_article_summary"], 34 | } 35 | 36 | llm = ChatOpenAI(temperature=0, model="gpt-4o-mini") 37 | 38 | urls = ["https://www.wsj.com"] 39 | 40 | extracted_content = scrape_with_playwright(urls, schema=schema) 41 | 42 | print("Extracted content:") 43 | print(extracted_content) 44 | -------------------------------------------------------------------------------- /notebooks/dev-notebooks/6.1-langchain-app-dev-structure.ipynb: -------------------------------------------------------------------------------- 1 | { 2 | "cells": [ 3 | { 4 | "cell_type": "code", 5 | "execution_count": null, 6 | "metadata": {}, 7 | "outputs": [], 8 | "source": [ 9 | "# uncomment and run below:\n", 10 | "%pip install langchain\n", 11 | "%pip install langchain-openai\n", 12 | "%pip install -U langsmith" 13 | ] 14 | }, 15 | { 16 | "cell_type": "markdown", 17 | "metadata": {}, 18 | "source": [ 19 | "1. Environment setup\n", 20 | "\n", 21 | "2. Prototype in with jupyter notebook + langsmith for inspection\n", 22 | "\n", 23 | "3. Set up the chain in the project app structure\n", 24 | "\n", 25 | "4. Set up server\n", 26 | "\n", 27 | "5. Deploy locally\n", 28 | "\n", 29 | "6. Test it on the playground\n", 30 | "\n", 31 | "7. Call it as an API" 32 | ] 33 | }, 34 | { 35 | "cell_type": "markdown", 36 | "metadata": {}, 37 | "source": [ 38 | "# 1. Environment setup\n", 39 | "\n", 40 | "```\n", 41 | "conda create -n rag-pdf-app python=3.11\n", 42 | "conda activate rag-pdf-app\n", 43 | "pip install -U \"langchain-cli [serve]\" \"langserve [all]\"\n", 44 | "langchain app new .\n", 45 | "poetry add langchain\n", 46 | "poetry add langchain-community\n", 47 | "poetry add langchain-openai\n", 48 | "poetry add chromadb\n", 49 | "poetry add pypdf\n", 50 | "poetry add tiktoken\n", 51 | "poetry add openai\n", 52 | "poetry add jupyter\n", 53 | "poetry add python-dotenv\n", 54 | "```\n", 55 | "\n", 56 | "Below is optional depending on your development setup:\n", 57 | "\n", 58 | "```\n", 59 | "poetry run jupyter notebook\n", 60 | "```" 61 | ] 62 | }, 63 | { 64 | "cell_type": "markdown", 65 | "metadata": {}, 66 | "source": [ 67 | "# 2. Prototype in with jupyter notebook + langsmith for inspection" 68 | ] 69 | }, 70 | { 71 | "cell_type": "code", 72 | "execution_count": 1, 73 | "metadata": {}, 74 | "outputs": [], 75 | "source": [ 76 | "import os\n", 77 | "import getpass\n", 78 | "\n", 79 | "os.environ[\"LANGCHAIN_TRACING_V2\"] = \"true\"\n", 80 | "os.environ[\"LANGCHAIN_API_KEY\"] = getpass.getpass()" 81 | ] 82 | }, 83 | { 84 | "cell_type": "code", 85 | "execution_count": null, 86 | "metadata": {}, 87 | "outputs": [], 88 | "source": [ 89 | "# in rag-pdf-app/app/chain.py\n", 90 | "\n", 91 | "# inspired by this template from langchain: https://github.com/langchain-ai/langchain/blob/master/templates/rag-chroma-private/rag_chroma_private/chain.py\n", 92 | "\n", 93 | "from langchain_community.document_loaders import PyPDFLoader\n", 94 | "from langchain_community.embeddings import OllamaEmbeddings\n", 95 | "from langchain_core.prompts import PromptTemplate, ChatPromptTemplate, MessagesPlaceholder\n", 96 | "from langchain.chains import RetrievalQA\n", 97 | "from langchain_core.output_parsers import StrOutputParser\n", 98 | "from langchain_core.runnables import RunnablePassthrough, RunnableParallel, RunnableLambda\n", 99 | "from langchain_community.vectorstores import Chroma\n", 100 | "from langchain_community.chat_models import ChatOllama\n", 101 | "from langchain_core.pydantic_v1 import BaseModel, Field\n", 102 | "from langchain import hub\n", 103 | "from typing import List, Tuple\n", 104 | "\n", 105 | "def load_pdf(file_path: str=\"./paper.pdf\") -> str:\n", 106 | " loader = PyPDFLoader(file_path)\n", 107 | " return loader.load()\n", 108 | "\n", 109 | "def index_docs(docs: List[str], \n", 110 | " persist_directory: str=\"./i-shall-persist\", \n", 111 | " embedding_model: str=\"llama3\"):\n", 112 | " embeddings = OllamaEmbeddings(model=embedding_model)\n", 113 | " vectordb = Chroma.from_documents(docs, embeddings, persist_directory=persist_directory)\n", 114 | " retriever = vectordb.as_retriever()\n", 115 | " \n", 116 | " return retriever\n", 117 | "\n", 118 | "\n", 119 | "file_path = \"./paper.pdf\"\n", 120 | "\n", 121 | "docs = load_pdf(file_path)\n", 122 | "\n", 123 | "retriever = index_docs(docs)\n", 124 | "\n", 125 | "template = \"\"\"\n", 126 | "Ansdwer the question based only on the following context:\n", 127 | "{context}\n", 128 | "\n", 129 | "Question: {question}\n", 130 | "\"\"\"\n", 131 | "\n", 132 | "prompt = ChatPromptTemplate.from_template(template)\n", 133 | "\n", 134 | "llm = ChatOllama(model=\"llama3\")\n", 135 | "\n", 136 | "# 2 suggestions for creating the rag chain:\n", 137 | "\n", 138 | "# chain = (\n", 139 | "# RunnableParallel({\"context\": retriever, \"question\": RunnablePassthrough()}) # RunnablePassthrough source: https://api.python.langchain.com/en/latest/runnables/langchain_core.runnables.passthrough.RunnablePassthrough.html#langchain-core-runnables-passthrough-runnablepassthrough:~:text=Runnable%20to%20passthrough,and%20experiment%20with.\n", 140 | "# # RunnableParallel source: https://api.python.langchain.com/en/latest/runnables/langchain_core.runnables.base.RunnableParallel.html\n", 141 | "# | prompt\n", 142 | "# | llm\n", 143 | "# | StrOutputParser()\n", 144 | "# )\n", 145 | "\n", 146 | "chain = RetrievalQA.from_chain_type(llm, retriever=retriever, chain_type_kwargs={\"prompt\": prompt}, return_source_documents=True) | RunnableLambda(lambda x: x[\"result\"])\n", 147 | "# qa_chain = RetrievalQA.from_chain_type(llm, retriever=retriever, return_source_documents=True)\n", 148 | "\n", 149 | "# Add typing for input\n", 150 | "class Question(BaseModel):\n", 151 | " __root__: str\n", 152 | " # The __root__ field in Pydantic models is used to define a model\n", 153 | " # where you expect a single value or a list rather than a dictionary \n", 154 | " # of named fields. Essentially, it allows your model to handle instances \n", 155 | " # where data does not naturally fit into a key-value structure, \n", 156 | " # such as a single value or a list.\n", 157 | "\n", 158 | "\n", 159 | "rag_chain = chain.with_types(input_type=Question)" 160 | ] 161 | }, 162 | { 163 | "cell_type": "markdown", 164 | "metadata": {}, 165 | "source": [ 166 | "# 3. Set up the chain in the project app structure\n", 167 | "\n", 168 | "![](./assets-resources/langchain-project-structure.png)" 169 | ] 170 | }, 171 | { 172 | "cell_type": "markdown", 173 | "metadata": {}, 174 | "source": [ 175 | "Go to the folder: `rag-pdf-app/app/`\n", 176 | "\n", 177 | "and save the rag code from above in `chain.py`" 178 | ] 179 | }, 180 | { 181 | "cell_type": "markdown", 182 | "metadata": {}, 183 | "source": [ 184 | "# 4. Set up server\n", 185 | "\n", 186 | "Go to `rag-pdf-app/app/server.py`\n", 187 | "\n", 188 | "and change:\n", 189 | "\n", 190 | "```python\n", 191 | "from fastapi import FastAPI\n", 192 | "from fastapi.responses import RedirectResponse\n", 193 | "from langserve import add_routes\n", 194 | "from app.chain import rag_chain\n", 195 | "\n", 196 | "app = FastAPI()\n", 197 | "\n", 198 | "\n", 199 | "@app.get(\"/\")\n", 200 | "async def redirect_root_to_docs():\n", 201 | " return RedirectResponse(\"/docs\")\n", 202 | "\n", 203 | "\n", 204 | "# Edit this to add the chain you want to add\n", 205 | "# Add routes connects the chain to our app exposing the methods of the chain to our web server\n", 206 | "add_routes(app,rag_chain, path=\"/rag-local\")\n", 207 | "\n", 208 | "if __name__ == \"__main__\":\n", 209 | " import uvicorn\n", 210 | "\n", 211 | " uvicorn.run(app, host=\"0.0.0.0\", port=8000)\n", 212 | "```" 213 | ] 214 | }, 215 | { 216 | "cell_type": "markdown", 217 | "metadata": {}, 218 | "source": [ 219 | "# 5. Deploy locally" 220 | ] 221 | }, 222 | { 223 | "cell_type": "markdown", 224 | "metadata": {}, 225 | "source": [ 226 | "From within the conda environment in the root folder of the project:\n", 227 | "\n", 228 | "`langchain serve`\n", 229 | "\n", 230 | "or with poetry:\n", 231 | "\n", 232 | "`poetry run langchain serve --port=8100`" 233 | ] 234 | }, 235 | { 236 | "cell_type": "markdown", 237 | "metadata": {}, 238 | "source": [ 239 | "# 6. Test it on the playground\n", 240 | "\n", 241 | "`rag-local/playground/`" 242 | ] 243 | }, 244 | { 245 | "cell_type": "markdown", 246 | "metadata": {}, 247 | "source": [] 248 | }, 249 | { 250 | "cell_type": "markdown", 251 | "metadata": {}, 252 | "source": [ 253 | "# 7. Call it as an API\n", 254 | "\n", 255 | "```\n", 256 | "from langserve.client import RemoteRunnable\n", 257 | "\n", 258 | "runnable = RemoteRunnable(\"http://localhost:8000/rag-local\")\n", 259 | "\n", 260 | "runnable.invoke(\"What is attention in this paper?\")\n", 261 | "```" 262 | ] 263 | } 264 | ], 265 | "metadata": { 266 | "kernelspec": { 267 | "display_name": "oreilly-langchain", 268 | "language": "python", 269 | "name": "oreilly-langchain" 270 | }, 271 | "language_info": { 272 | "codemirror_mode": { 273 | "name": "ipython", 274 | "version": 3 275 | }, 276 | "file_extension": ".py", 277 | "mimetype": "text/x-python", 278 | "name": "python", 279 | "nbconvert_exporter": "python", 280 | "pygments_lexer": "ipython3", 281 | "version": "3.11.9" 282 | } 283 | }, 284 | "nbformat": 4, 285 | "nbformat_minor": 2 286 | } 287 | -------------------------------------------------------------------------------- /notebooks/dev-notebooks/create-knowledge-graph.ipynb: -------------------------------------------------------------------------------- 1 | { 2 | "cells": [ 3 | { 4 | "cell_type": "code", 5 | "execution_count": 3, 6 | "metadata": {}, 7 | "outputs": [ 8 | { 9 | "data": { 10 | "image/svg+xml": [ 11 | "\n", 12 | "\n", 14 | "\n", 16 | "\n", 17 | "\n", 19 | "\n", 20 | "\n", 21 | "\n", 22 | "\n", 23 | "1\n", 24 | "\n", 25 | "Key Decisions\n", 26 | "\n", 27 | "\n", 28 | "\n", 29 | "2\n", 30 | "\n", 31 | "Follow-up Items\n", 32 | "\n", 33 | "\n", 34 | "\n", 35 | "1->2\n", 36 | "\n", 37 | "\n", 38 | "Decisions lead to Follow-up Items\n", 39 | "\n", 40 | "\n", 41 | "\n", 42 | "3\n", 43 | "\n", 44 | "Associated Owners\n", 45 | "\n", 46 | "\n", 47 | "\n", 48 | "2->3\n", 49 | "\n", 50 | "\n", 51 | "Follow-up Items assigned to Owners\n", 52 | "\n", 53 | "\n", 54 | "\n", 55 | "3->1\n", 56 | "\n", 57 | "\n", 58 | "Owners influence Key Decisions\n", 59 | "\n", 60 | "\n", 61 | "\n" 62 | ], 63 | "text/plain": [ 64 | "" 65 | ] 66 | }, 67 | "metadata": {}, 68 | "output_type": "display_data" 69 | } 70 | ], 71 | "source": [ 72 | "from pydantic import BaseModel, Field\n", 73 | "from typing import List\n", 74 | "from openai import OpenAI\n", 75 | "import instructor\n", 76 | "from graphviz import Digraph\n", 77 | "import argparse\n", 78 | "\n", 79 | "class Node(BaseModel):\n", 80 | " id: int\n", 81 | " label: str\n", 82 | " color: str\n", 83 | "\n", 84 | "class Edge(BaseModel):\n", 85 | " source: int\n", 86 | " target: int\n", 87 | " label: str\n", 88 | " color: str = \"black\"\n", 89 | "\n", 90 | "class KnowledgeGraph(BaseModel):\n", 91 | " nodes: List[Node] = Field(..., default_factory=list)\n", 92 | " edges: List[Edge] = Field(..., default_factory=list)\n", 93 | "\n", 94 | "# Adds response_model to ChatCompletion\n", 95 | "# Allows the return of Pydantic model rather than raw JSON\n", 96 | "client = instructor.patch(OpenAI())\n", 97 | "\n", 98 | "def generate_graph(input) -> KnowledgeGraph:\n", 99 | " return client.chat.completions.create(\n", 100 | " model=\"gpt-3.5-turbo\",\n", 101 | " messages=[\n", 102 | " {\n", 103 | " \"role\": \"user\",\n", 104 | " \"content\": f\"Help me understand the following by describing it as a detailed knowledge graph: {input}\",\n", 105 | " }\n", 106 | " ],\n", 107 | " response_model=KnowledgeGraph,\n", 108 | " ) # type: ignore\n", 109 | "\n", 110 | "def visualize_knowledge_graph(kg: KnowledgeGraph):\n", 111 | " dot = Digraph(comment=\"Knowledge Graph\")\n", 112 | "\n", 113 | " # Add nodes\n", 114 | " for node in kg.nodes:\n", 115 | " dot.node(str(node.id), node.label, color=node.color)\n", 116 | "\n", 117 | " # Add edges\n", 118 | " for edge in kg.edges:\n", 119 | " dot.edge(str(edge.source), str(edge.target), label=edge.label, color=edge.color)\n", 120 | "\n", 121 | " # Render the graph\n", 122 | " # dot.render(\"knowledge_graph.gv\", view=True)\n", 123 | " display(dot)\n", 124 | " \n", 125 | " \n", 126 | "\n", 127 | "graph: KnowledgeGraph = generate_graph(\"\"\"\n", 128 | "First, list the key decisions, follow-up items, and associated owners in a sketchpad.\n", 129 | "Then, check that the details in the sketchpad are factually consistent with the transcript.\n", 130 | "Finally, synthesize the key points into a concise summary.\"\"\")\n", 131 | "visualize_knowledge_graph(graph)" 132 | ] 133 | } 134 | ], 135 | "metadata": { 136 | "kernelspec": { 137 | "display_name": "automations", 138 | "language": "python", 139 | "name": "automations" 140 | }, 141 | "language_info": { 142 | "codemirror_mode": { 143 | "name": "ipython", 144 | "version": 3 145 | }, 146 | "file_extension": ".py", 147 | "mimetype": "text/x-python", 148 | "name": "python", 149 | "nbconvert_exporter": "python", 150 | "pygments_lexer": "ipython3", 151 | "version": "3.11.9" 152 | } 153 | }, 154 | "nbformat": 4, 155 | "nbformat_minor": 2 156 | } 157 | -------------------------------------------------------------------------------- /notebooks/dev-notebooks/knowledge_graph.gv: -------------------------------------------------------------------------------- 1 | // Knowledge Graph 2 | digraph { 3 | 1 [label="Random Variable" color=lightblue] 4 | 2 [label="Probability Distribution" color=lightblue] 5 | 3 [label="Joint Probability Mass Function" color=lightgreen] 6 | 4 [label="Discrete Random Variables" color=lightblue] 7 | 5 [label="Probability Mass Function" color=lightgreen] 8 | 1 -> 2 [label=has color=black] 9 | 2 -> 3 [label=has color=black] 10 | 1 -> 4 [label="is a" color=black] 11 | 4 -> 5 [label=has color=black] 12 | } 13 | -------------------------------------------------------------------------------- /notebooks/dev-notebooks/knowledge_graph.gv.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/dev-notebooks/knowledge_graph.gv.pdf -------------------------------------------------------------------------------- /notebooks/dev-notebooks/llama3-langchain-basic-chain.ipynb: -------------------------------------------------------------------------------- 1 | { 2 | "cells": [ 3 | { 4 | "cell_type": "markdown", 5 | "metadata": {}, 6 | "source": [ 7 | "# Basic Chain Example LangChain + Llama3" 8 | ] 9 | }, 10 | { 11 | "cell_type": "code", 12 | "execution_count": 2, 13 | "metadata": {}, 14 | "outputs": [], 15 | "source": [ 16 | "from langchain_community.chat_models import ChatOllama\n", 17 | "from langchain_core.output_parsers import StrOutputParser\n", 18 | "from langchain_core.prompts import ChatPromptTemplate" 19 | ] 20 | }, 21 | { 22 | "cell_type": "code", 23 | "execution_count": 3, 24 | "metadata": {}, 25 | "outputs": [], 26 | "source": [ 27 | "\n", 28 | "# Local Llama3\n", 29 | "llm = ChatOllama(\n", 30 | " model=\"llama3\",\n", 31 | " keep_alive=-1, # keep the model loaded indefinitely\n", 32 | " temperature=0,\n", 33 | " max_new_tokens=512\n", 34 | ")" 35 | ] 36 | }, 37 | { 38 | "cell_type": "code", 39 | "execution_count": 4, 40 | "metadata": {}, 41 | "outputs": [], 42 | "source": [ 43 | "prompt = ChatPromptTemplate.from_template(\"Write me a 500 word article on {topic} from the perspective of a {profession}.\")" 44 | ] 45 | }, 46 | { 47 | "cell_type": "code", 48 | "execution_count": 7, 49 | "metadata": {}, 50 | "outputs": [ 51 | { 52 | "name": "stdout", 53 | "output_type": "stream", 54 | "text": [ 55 | "**\"LLMs: The Game-Changers for Shipping Magnates Like Me\"**\n", 56 | "\n", 57 | "As a shipping magnate, I've spent my fair share of time navigating the complexities of global trade and logistics. From optimizing routes to managing inventory, every decision counts when it comes to keeping my fleet of vessels running smoothly and efficiently. That's why I'm excited about the potential of Large Language Models (LLMs) to revolutionize our industry.\n", 58 | "\n", 59 | "In recent years, LLMs have made headlines for their impressive language processing capabilities. These AI-powered models can understand and generate human-like text with uncanny accuracy. But what does this mean for shipping magnates like me? Let me tell you – it's a game-changer.\n", 60 | "\n", 61 | "Firstly, LLMs can help streamline communication between our teams and clients. Imagine being able to quickly generate responses to common customer inquiries or providing detailed reports on shipment status without breaking a sweat. With an LLM at your disposal, you can automate routine tasks and free up your team to focus on higher-value activities like strategic planning and problem-solving.\n", 62 | "\n", 63 | "But the benefits don't stop there. LLMs can also be used to analyze vast amounts of data related to shipping patterns, market trends, and regulatory compliance. By processing this information in real-time, we can identify opportunities for optimization and make more informed decisions about our operations. For instance, an LLM could help us predict demand fluctuations based on historical data and adjust our logistics accordingly.\n", 64 | "\n", 65 | "Another area where LLMs can make a significant impact is in the realm of documentation and compliance. Shipping companies are subject to numerous regulations and standards, which can be time-consuming and error-prone to navigate. An LLM can assist with generating accurate and compliant documents, such as customs forms and insurance claims, reducing the risk of errors and fines.\n", 66 | "\n", 67 | "Furthermore, LLMs have the potential to revolutionize our approach to supply chain management. By analyzing vast amounts of data on shipment routes, modes of transportation, and inventory levels, an LLM can help us identify areas for improvement and optimize our logistics networks. This could lead to significant cost savings, reduced transit times, and improved customer satisfaction.\n", 68 | "\n", 69 | "Of course, there are also concerns about the potential risks associated with relying on AI-powered models like LLMs. As a shipping magnate, I understand the importance of maintaining human oversight and control over critical decision-making processes. However, I believe that LLMs can be designed to augment our existing capabilities rather than replace them.\n", 70 | "\n", 71 | "In conclusion, Large Language Models have the potential to transform the shipping industry in numerous ways. By streamlining communication, analyzing data, generating documents, and optimizing supply chains, LLMs can help us operate more efficiently, effectively, and profitably. As a shipping magnate, I'm excited about the possibilities that these AI-powered models bring to our industry – and I'm eager to see how they will shape the future of global trade and logistics.\n", 72 | "\n", 73 | "**About the Author**\n", 74 | "\n", 75 | "John Smith is a seasoned shipping magnate with over two decades of experience in the industry. He has built his reputation on his ability to navigate complex logistical challenges and optimize supply chain operations. In this article, he shares his insights on the potential of Large Language Models to revolutionize the shipping industry.\n" 76 | ] 77 | } 78 | ], 79 | "source": [ 80 | "# using LangChain Expressive Language chain syntax\n", 81 | "chain = prompt | llm | StrOutputParser()" 82 | ] 83 | }, 84 | { 85 | "cell_type": "code", 86 | "execution_count": null, 87 | "metadata": {}, 88 | "outputs": [], 89 | "source": [ 90 | "print(chain.invoke({\"topic\": \"LLMs\", \"profession\": \"shipping magnate\"}))" 91 | ] 92 | } 93 | ], 94 | "metadata": { 95 | "kernelspec": { 96 | "display_name": "oreilly-llama3", 97 | "language": "python", 98 | "name": "oreilly-llama3" 99 | }, 100 | "language_info": { 101 | "codemirror_mode": { 102 | "name": "ipython", 103 | "version": 3 104 | }, 105 | "file_extension": ".py", 106 | "mimetype": "text/x-python", 107 | "name": "python", 108 | "nbconvert_exporter": "python", 109 | "pygments_lexer": "ipython3", 110 | "version": "3.11.9" 111 | } 112 | }, 113 | "nbformat": 4, 114 | "nbformat_minor": 2 115 | } 116 | -------------------------------------------------------------------------------- /notebooks/dev-notebooks/research-assistant_test.py: -------------------------------------------------------------------------------- 1 | from langchain.chat_models import ChatOpenAI 2 | from langchain.prompts import ChatPromptTemplate 3 | from langchain.schema.output_parser import StrOutputParser 4 | import requests 5 | from bs4 import BeautifulSoup 6 | from langchain.schema.runnable import RunnablePassthrough, RunnableLambda 7 | from langchain.utilities import DuckDuckGoSearchAPIWrapper 8 | import json 9 | 10 | RESULTS_PER_QUESTION = 3 11 | 12 | ddg_search = DuckDuckGoSearchAPIWrapper() 13 | 14 | 15 | def web_search(query: str, num_results: int = RESULTS_PER_QUESTION): 16 | results = ddg_search.results(query, num_results) 17 | return [r["link"] for r in results] 18 | 19 | 20 | SUMMARY_TEMPLATE = """{text} 21 | ----------- 22 | Using the above text, answer in short the following question: 23 | > {question} 24 | ----------- 25 | if the question cannot be answered using the text, imply summarize the text. Include all factual information, numbers, stats etc if available.""" # noqa: E501 26 | SUMMARY_PROMPT = ChatPromptTemplate.from_template(SUMMARY_TEMPLATE) 27 | 28 | 29 | def scrape_text(url: str): 30 | # Send a GET request to the webpage 31 | try: 32 | response = requests.get(url) 33 | 34 | # Check if the request was successful 35 | if response.status_code == 200: 36 | # Parse the content of the request with BeautifulSoup 37 | soup = BeautifulSoup(response.text, "html.parser") 38 | 39 | # Extract all text from the webpage 40 | page_text = soup.get_text(separator=" ", strip=True) 41 | 42 | # Print the extracted text 43 | return page_text 44 | else: 45 | return f"Failed to retrieve the webpage: Status code {response.status_code}" 46 | except Exception as e: 47 | print(e) 48 | return f"Failed to retrieve the webpage: {e}" 49 | 50 | 51 | url = "https://blog.langchain.dev/announcing-langsmith/" 52 | 53 | scrape_and_summarize_chain = RunnablePassthrough.assign( 54 | summary = RunnablePassthrough.assign( 55 | text=lambda x: scrape_text(x["url"])[:10000] 56 | ) | SUMMARY_PROMPT | ChatOpenAI(model="gpt-3.5-turbo-1106") | StrOutputParser() 57 | ) | (lambda x: f"URL: {x['url']}\n\nSUMMARY: {x['summary']}") 58 | 59 | web_search_chain = RunnablePassthrough.assign( 60 | urls = lambda x: web_search(x["question"]) 61 | ) | (lambda x: [{"question": x["question"], "url": u} for u in x["urls"]]) | scrape_and_summarize_chain.map() 62 | 63 | 64 | SEARCH_PROMPT = ChatPromptTemplate.from_messages( 65 | [ 66 | ( 67 | "user", 68 | "Write 3 google search queries to search online that form an " 69 | "objective opinion from the following: {question}\n" 70 | "You must respond with a list of strings in the following format: " 71 | '["query 1", "query 2", "query 3"].', 72 | ), 73 | ] 74 | ) 75 | 76 | search_question_chain = SEARCH_PROMPT | ChatOpenAI(temperature=0) | StrOutputParser() | json.loads 77 | 78 | full_research_chain = search_question_chain | (lambda x: [{"question": q} for q in x]) | web_search_chain.map() 79 | 80 | WRITER_SYSTEM_PROMPT = "You are an AI critical thinker research assistant. Your sole purpose is to write well written, critically acclaimed, objective and structured reports on given text." # noqa: E501 81 | 82 | 83 | # Report prompts from https://github.com/assafelovic/gpt-researcher/blob/master/gpt_researcher/master/prompts.py 84 | RESEARCH_REPORT_TEMPLATE = """Information: 85 | -------- 86 | {research_summary} 87 | -------- 88 | Using the above information, answer the following question or topic: "{question}" in a detailed report -- \ 89 | The report should focus on the answer to the question, should be well structured, informative, \ 90 | in depth, with facts and numbers if available and a minimum of 1,200 words. 91 | You should strive to write the report as long as you can using all relevant and necessary information provided. 92 | You must write the report with markdown syntax. 93 | You MUST determine your own concrete and valid opinion based on the given information. Do NOT deter to general and meaningless conclusions. 94 | Write all used source urls at the end of the report, and make sure to not add duplicated sources, but only one reference for each. 95 | You must write the report in apa format. 96 | Please do your best, this is very important to my career.""" # noqa: E501 97 | 98 | prompt = ChatPromptTemplate.from_messages( 99 | [ 100 | ("system", WRITER_SYSTEM_PROMPT), 101 | ("user", RESEARCH_REPORT_TEMPLATE), 102 | ] 103 | ) 104 | 105 | def collapse_list_of_lists(list_of_lists): 106 | content = [] 107 | for l in list_of_lists: 108 | content.append("\n\n".join(l)) 109 | return "\n\n".join(content) 110 | 111 | chain = RunnablePassthrough.assign( 112 | research_summary= full_research_chain | collapse_list_of_lists 113 | ) | prompt | ChatOpenAI(model="gpt-3.5-turbo-1106") | StrOutputParser() 114 | 115 | #!/usr/bin/env python 116 | # from fastapi import FastAPI 117 | # from langserve import add_routes 118 | 119 | # app = FastAPI( 120 | # title="LangChain Server", 121 | # version="1.0", 122 | # description="A simple api server using Langchain's Runnable interfaces", 123 | # ) 124 | 125 | # add_routes( 126 | # app, 127 | # chain, 128 | # path="/research-assistant", 129 | # ) 130 | 131 | 132 | # if __name__ == "__main__": 133 | # import uvicorn 134 | 135 | # uvicorn.run(app, host="localhost", port=8000) 136 | -------------------------------------------------------------------------------- /notebooks/jira-agent.py: -------------------------------------------------------------------------------- 1 | from jira import JIRA 2 | import argparse 3 | import os 4 | from langchain.tools import tool 5 | from langchain.agents import initialize_agent, AgentExecutor 6 | from langchain.prompts import ChatPromptTemplate, MessagesPlaceholder 7 | from langchain.tools.render import format_tool_to_openai_function 8 | from langchain.chat_models import ChatOpenAI 9 | from langchain.agents.format_scratchpad import format_to_openai_function_messages 10 | from langchain.agents.output_parsers import OpenAIFunctionsAgentOutputParser 11 | import sys 12 | 13 | 14 | @tool 15 | def list_issue_transition_options(issue_key,server_url=''): 16 | """Lists all the available transition options for a Jira issue given the issue key""" 17 | JIRA_USERNAME = os.environ["JIRA_USERNAME"] 18 | JIRA_TOKEN = os.environ["JIRA_TOKEN"] 19 | 20 | jira = JIRA(basic_auth=(JIRA_USERNAME, JIRA_TOKEN), server=server_url) 21 | transitions = jira.transitions(issue_key) 22 | for transition in transitions: 23 | print(transition['name'], transition['id']) 24 | 25 | @tool 26 | def update_issue_status(issue_key, target_status_name, server_url=''): 27 | """Updates the status of a Jira issue given the issue key and the target status name""" 28 | 29 | JIRA_USERNAME = os.environ["JIRA_USERNAME"] 30 | JIRA_TOKEN = os.environ["JIRA_TOKEN"] 31 | 32 | jira = JIRA(basic_auth=(JIRA_USERNAME, JIRA_TOKEN), server=server_url) 33 | transitions = jira.transitions(issue_key) 34 | target_transition_id = None 35 | 36 | # Find the transition ID for the target status 37 | for transition in transitions: 38 | print(transition['name'].lower().strip()) 39 | if transition['name'].lower().strip() == target_status_name.lower(): 40 | target_transition_id = transition['id'] 41 | break 42 | 43 | # Execute the transition if possible 44 | if target_transition_id: 45 | jira.transition_issue(issue_key, target_transition_id) 46 | print(f"Issue {issue_key} has been moved to '{target_status_name}'.") 47 | else: 48 | print(f"Transition to '{target_status_name}' not found.") 49 | 50 | @tool 51 | def create_issue(summary, description, issue_type, project_key='ML', server_url=''): 52 | """Creates a Jira issue with summary, description, issue type and a project key""" 53 | JIRA_USERNAME = os.environ["JIRA_USERNAME"] 54 | JIRA_TOKEN = os.environ["JIRA_TOKEN"] 55 | 56 | jira = JIRA(basic_auth=(JIRA_USERNAME, JIRA_TOKEN), server=server_url) 57 | issue_dict = { 58 | 'project': {'key': project_key}, 59 | 'summary': summary, 60 | 'description': description, 61 | 'issuetype': {'name': issue_type}, 62 | 63 | } 64 | new_issue = jira.create_issue(fields=issue_dict) 65 | print(f'New issue created with key: {new_issue.key}') 66 | 67 | 68 | @tool 69 | def delete_issue(issue_key, server_url=''): 70 | """Deletes a Jira issue given the issue key""" 71 | JIRA_USERNAME = os.environ["JIRA_USERNAME"] 72 | JIRA_TOKEN = os.environ["JIRA_TOKEN"] 73 | jira = JIRA(basic_auth=(JIRA_USERNAME, JIRA_TOKEN), server=server_url) 74 | issue = jira.issue(issue_key) 75 | print(f'Deleting issue: {issue.key}') 76 | delete_issue = input("Do you want to delete the issue? (y/n): ") 77 | if delete_issue.lower() in ['y', 'yes']: 78 | issue.delete() 79 | print('Issue deleted successfully') 80 | 81 | 82 | @tool 83 | def update_issue_summary(issue_key, summary, server_url=''): 84 | """Updates issue summary""" 85 | JIRA_USERNAME = os.environ["JIRA_USERNAME"] 86 | JIRA_TOKEN = os.environ["JIRA_TOKEN"] 87 | 88 | jira = JIRA(basic_auth=(JIRA_USERNAME, JIRA_TOKEN), server=server_url) 89 | issue = jira.issue(issue_key) 90 | issue.update(summary=summary) 91 | print(f'Issue {issue.key} summary updated successfully') 92 | 93 | 94 | @tool 95 | def update_issue_description(issue_key, description, server_url=''): 96 | """Updates issue description""" 97 | JIRA_USERNAME = os.environ["JIRA_USERNAME"] 98 | JIRA_TOKEN = os.environ["JIRA_TOKEN"] 99 | 100 | jira = JIRA(basic_auth=(JIRA_USERNAME, JIRA_TOKEN), server=server_url) 101 | issue = jira.issue(issue_key) 102 | issue.update(description=description) 103 | print(f'Issue {issue.key} description updated successfully') 104 | 105 | @tool 106 | def view_issue(issue_key, server_url=''): 107 | """Views a Jira issue given the issue key""" 108 | JIRA_USERNAME = os.environ["JIRA_USERNAME"] 109 | JIRA_TOKEN = os.environ["JIRA_TOKEN"] 110 | 111 | jira = JIRA(basic_auth=(JIRA_USERNAME, JIRA_TOKEN), server=server_url) 112 | """Views a Jira issue given the issue key""" 113 | issue = jira.issue(issue_key) 114 | print(f'Viewing issue {issue.key}.') 115 | 116 | 117 | def setup_prompt_template(): 118 | prompt = ChatPromptTemplate.from_messages( 119 | [ 120 | ( 121 | "system", 122 | "You are very powerful assistant that helps users to manage their issues in the Jira Software.", 123 | ), 124 | ("user", "{input}"), 125 | MessagesPlaceholder(variable_name="agent_scratchpad"), 126 | ]) 127 | 128 | return prompt 129 | 130 | 131 | def setup_agent(prompt, llm_with_tools): 132 | agent = ( 133 | { 134 | "input": lambda x: x["input"], 135 | "agent_scratchpad": lambda x: format_to_openai_function_messages( 136 | x["intermediate_steps"] 137 | ), 138 | } 139 | | prompt 140 | | llm_with_tools 141 | | OpenAIFunctionsAgentOutputParser()) 142 | agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True) 143 | 144 | return agent_executor 145 | 146 | if __name__=="__main__": 147 | action_input = sys.argv[1] 148 | prompt = setup_prompt_template() 149 | llm = ChatOpenAI() 150 | tools = [ 151 | view_issue, 152 | create_issue, 153 | update_issue_summary, 154 | update_issue_description, 155 | delete_issue, 156 | update_issue_status, 157 | list_issue_transition_options, 158 | ] 159 | llm_with_tools = llm.bind(functions=[format_tool_to_openai_function(t) for t in tools]) 160 | agent_executor = setup_agent(prompt, llm_with_tools) 161 | agent_executor.invoke({"input": action_input}) 162 | -------------------------------------------------------------------------------- /notebooks/langchain-app.py: -------------------------------------------------------------------------------- 1 | # source: https://python.langchain.com/docs/tutorials/llm_chain/#:~:text=Server%E2%80%8B,with%20langserve.add_routes 2 | #!/usr/bin/env python 3 | from fastapi import FastAPI 4 | from langchain_core.prompts import ChatPromptTemplate 5 | from langchain_core.output_parsers import StrOutputParser 6 | from langchain_openai import ChatOpenAI 7 | from langserve import add_routes 8 | 9 | # 1. Create prompt template 10 | system_template = "Translate the following into {language}:" 11 | prompt_template = ChatPromptTemplate.from_messages([ 12 | ('system', system_template), 13 | ('user', '{text}') 14 | ]) 15 | 16 | # 2. Create model 17 | model = ChatOpenAI(model="gpt-4o-mini", temperature=0) 18 | 19 | # 3. Create parser 20 | parser = StrOutputParser() 21 | 22 | # 4. Create chain 23 | chain = prompt_template | model | parser 24 | 25 | 26 | # 4. App definition 27 | app = FastAPI( 28 | title="LangChain Server", 29 | version="1.0", 30 | description="A simple API server using LangChain's Runnable interfaces", 31 | ) 32 | 33 | # 5. Adding chain route 34 | add_routes( 35 | app, 36 | chain, 37 | path="/chain", 38 | ) 39 | 40 | if __name__ == "__main__": 41 | import uvicorn 42 | 43 | uvicorn.run(app, host="localhost", port=8000) -------------------------------------------------------------------------------- /notebooks/langchain-lcel-cheatsheet.md: -------------------------------------------------------------------------------- 1 | # LangChain Expression Language (LCEL) Cheat Sheet 2 | 3 | - **Invoke a runnable:** Use `Runnable.invoke()` for synchronous and `Runnable.ainvoke()` for asynchronous invocation. 4 | - **Batch a runnable:** Use `Runnable.batch()` for synchronous and `Runnable.abatch()` for asynchronous batching. 5 | - **Stream a runnable:** Use `Runnable.stream()` for synchronous and `Runnable.astream()` for asynchronous streaming. 6 | - **Compose runnables:** Use the pipe operator `|` to chain runnables. 7 | - **Invoke runnables in parallel:** Use `RunnableParallel` to run multiple runnables concurrently. 8 | - **Turn any function into a runnable:** Use `RunnableLambda` to convert a function into a runnable. 9 | - **Merge input and output dicts:** Use `RunnablePassthrough.assign` to merge input and output dictionaries. 10 | - **Include input dict in output dict:** Use `RunnablePassthrough` to include input in the output. 11 | - **Add default invocation args:** Use `Runnable.bind` to bind default arguments to a runnable. 12 | - **Add fallbacks:** Use `Runnable.with_fallbacks` to add fallback runnables. 13 | - **Add retries:** Use `Runnable.with_retry` to add retry logic to a runnable. 14 | - **Configure runnable execution:** Use `RunnableConfig` to specify execution configurations. 15 | - **Add default config to runnable:** Use `Runnable.with_config` to add default configurations. 16 | - **Make runnable attributes configurable:** Use `Runnable.with_configurable_fields` for configurable attributes. 17 | - **Make chain components configurable:** Use `Runnable.with_configurable_alternatives` for configurable chain components. 18 | - **Build a chain dynamically based on input:** Use conditionals to dynamically build chains. 19 | - **Generate a stream of events:** Use `Runnable.astream_events` for asynchronous event streams. 20 | - **Yield batched outputs as they complete:** Use `Runnable.batch_as_completed` for synchronous and `Runnable.abatch_as_completed` for asynchronous batching. 21 | - **Return subset of output dict:** Use `Runnable.pick` to select specific parts of the output. 22 | - **Declaratively make a batched version of a runnable:** Use `Runnable.map` for batching. 23 | - **Get a graph representation of a runnable:** Use `Runnable.get_graph` to visualize the runnable. 24 | - **Get all prompts in a chain:** Use `Runnable.get_prompts` to retrieve prompts from a chain. 25 | - **Add lifecycle listeners:** Use `Runnable.with_listeners` to add start and end listeners to a runnable. -------------------------------------------------------------------------------- /notebooks/langchain-structured-output-ui.py: -------------------------------------------------------------------------------- 1 | import streamlit as st 2 | from langchain_openai import ChatOpenAI 3 | from pydantic import BaseModel, Field 4 | import pandas as pd 5 | 6 | 7 | if 'df' not in st.session_state: 8 | st.session_state.df = pd.DataFrame(columns=["Product Name", "Price", "Description"]) 9 | 10 | class Product(BaseModel): 11 | name: str = Field(description="The name of the product") 12 | price: float = Field(description="The price of the product") 13 | description: str = Field(description="A detailed description of the product") 14 | 15 | llm = ChatOpenAI(model="gpt-4o-mini", temperature=0) 16 | 17 | llm_structured_output = llm.with_structured_output(Product) 18 | 19 | 20 | st.title("LangChain Structured Output UI") 21 | 22 | col1, col2 = st.columns(2) 23 | 24 | 25 | raw_text = st.text_area("Raw Text") 26 | 27 | # Add a checkbox to show the current table 28 | show_table = st.checkbox("Show current table") 29 | 30 | # Display the current table if the checkbox is checked 31 | if show_table: 32 | st.write("Current Product Information") 33 | st.table(st.session_state.df) 34 | 35 | if raw_text: 36 | st.write("Product Information") 37 | if raw_text!="" and st.button("Parse"): 38 | product_info = llm_structured_output.invoke(raw_text) 39 | st.write("Product Name,Price,Description") 40 | new_row = pd.DataFrame({ 41 | "Product Name": [product_info.name], 42 | "Price": [product_info.price], 43 | "Description": [product_info.description] 44 | }) 45 | st.session_state.df = pd.concat([st.session_state.df, new_row], ignore_index=True) 46 | st.table(st.session_state.df) 47 | else: 48 | st.write("Enter text in the left column to see it here.") -------------------------------------------------------------------------------- /notebooks/paper.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/notebooks/paper.pdf -------------------------------------------------------------------------------- /notebooks/rag_methods.py: -------------------------------------------------------------------------------- 1 | import os 2 | import dotenv 3 | from time import time 4 | import streamlit as st 5 | 6 | from langchain_community.document_loaders.text import TextLoader 7 | from langchain_community.document_loaders import ( 8 | WebBaseLoader, 9 | PyPDFLoader, 10 | Docx2txtLoader, 11 | CSVLoader, 12 | ) 13 | # pip install docx2txt, pypdf 14 | from langchain_community.vectorstores import Chroma 15 | from langchain.text_splitter import RecursiveCharacterTextSplitter 16 | from langchain_openai import OpenAIEmbeddings, AzureOpenAIEmbeddings 17 | from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder 18 | from langchain.chains import create_history_aware_retriever, create_retrieval_chain 19 | from langchain.chains.combine_documents import create_stuff_documents_chain 20 | 21 | dotenv.load_dotenv() 22 | 23 | os.environ["USER_AGENT"] = "myagent" 24 | DB_DOCS_LIMIT = 10 25 | 26 | # Function to stream the response of the LLM 27 | def stream_llm_response(llm_stream, messages): 28 | response_message = "" 29 | 30 | for chunk in llm_stream.stream(messages): 31 | response_message += chunk.content 32 | yield chunk 33 | 34 | st.session_state.messages.append({"role": "assistant", "content": response_message}) 35 | 36 | 37 | # --- Indexing Phase --- 38 | 39 | def load_doc_to_db(): 40 | # Use loader according to doc type 41 | if "rag_docs" in st.session_state and st.session_state.rag_docs: 42 | docs = [] 43 | for doc_file in st.session_state.rag_docs: 44 | if doc_file.name not in st.session_state.rag_sources: 45 | if len(st.session_state.rag_sources) < DB_DOCS_LIMIT: 46 | os.makedirs("source_files", exist_ok=True) 47 | file_path = f"./source_files/{doc_file.name}" 48 | with open(file_path, "wb") as file: 49 | file.write(doc_file.read()) 50 | 51 | try: 52 | if doc_file.type == "application/pdf": 53 | loader = PyPDFLoader(file_path) 54 | elif doc_file.name.endswith(".docx"): 55 | loader = Docx2txtLoader(file_path) 56 | elif doc_file.type in ["text/plain", "text/markdown"]: 57 | loader = TextLoader(file_path) 58 | elif doc_file.name.endswith(".csv"): 59 | loader = CSVLoader(file_path) 60 | else: 61 | st.warning(f"Document type {doc_file.type} not supported.") 62 | continue 63 | 64 | docs.extend(loader.load()) 65 | st.session_state.rag_sources.append(doc_file.name) 66 | 67 | except Exception as e: 68 | st.toast(f"Error loading document {doc_file.name}: {e}", icon="⚠️") 69 | print(f"Error loading document {doc_file.name}: {e}") 70 | 71 | finally: 72 | os.remove(file_path) 73 | 74 | else: 75 | st.error(F"Maximum number of documents reached ({DB_DOCS_LIMIT}).") 76 | 77 | if docs: 78 | _split_and_load_docs(docs) 79 | st.toast(f"Document *{str([doc_file.name for doc_file in st.session_state.rag_docs])[1:-1]}* loaded successfully.", icon="✅") 80 | 81 | 82 | def load_url_to_db(): 83 | if "rag_url" in st.session_state and st.session_state.rag_url: 84 | url = st.session_state.rag_url 85 | docs = [] 86 | if url not in st.session_state.rag_sources: 87 | if len(st.session_state.rag_sources) < 10: 88 | try: 89 | loader = WebBaseLoader(url) 90 | docs.extend(loader.load()) 91 | st.session_state.rag_sources.append(url) 92 | 93 | except Exception as e: 94 | st.error(f"Error loading document from {url}: {e}") 95 | 96 | if docs: 97 | _split_and_load_docs(docs) 98 | st.toast(f"Document from URL *{url}* loaded successfully.", icon="✅") 99 | 100 | else: 101 | st.error("Maximum number of documents reached (10).") 102 | 103 | 104 | def initialize_vector_db(docs): 105 | embedding = OpenAIEmbeddings(api_key=st.session_state.openai_api_key) 106 | 107 | # Create a persistent directory for the database 108 | persist_directory = f"./chroma_db_{st.session_state['session_id']}" 109 | 110 | vector_db = Chroma.from_documents( 111 | documents=docs, 112 | embedding=embedding, 113 | persist_directory=persist_directory, 114 | ) 115 | 116 | # Persist the database 117 | vector_db.persist() 118 | 119 | return vector_db 120 | 121 | 122 | def _split_and_load_docs(docs): 123 | text_splitter = RecursiveCharacterTextSplitter( 124 | chunk_size=1000, 125 | chunk_overlap=200, 126 | ) 127 | 128 | document_chunks = text_splitter.split_documents(docs) 129 | 130 | persist_directory = f"./chroma_db_{st.session_state['session_id']}" 131 | 132 | if os.path.exists(persist_directory): 133 | # Load existing database 134 | embedding = OpenAIEmbeddings(api_key=st.session_state.openai_api_key) 135 | st.session_state.vector_db = Chroma(persist_directory=persist_directory, embedding_function=embedding) 136 | # Add new documents to existing database 137 | st.session_state.vector_db.add_documents(document_chunks) 138 | else: 139 | # Create new database 140 | st.session_state.vector_db = initialize_vector_db(document_chunks) 141 | 142 | 143 | # --- Retrieval Augmented Generation (RAG) Phase --- 144 | 145 | def _get_context_retriever_chain(vector_db, llm): 146 | retriever = vector_db.as_retriever() 147 | prompt = ChatPromptTemplate.from_messages([ 148 | MessagesPlaceholder(variable_name="messages"), 149 | ("user", "{input}"), 150 | ("user", "Given the above conversation, generate a search query to look up in order to get inforamtion relevant to the conversation, focusing on the most recent messages."), 151 | ]) 152 | retriever_chain = create_history_aware_retriever(llm, retriever, prompt) 153 | 154 | return retriever_chain 155 | 156 | 157 | def get_conversational_rag_chain(llm): 158 | retriever_chain = _get_context_retriever_chain(st.session_state.vector_db, llm) 159 | 160 | prompt = ChatPromptTemplate.from_messages([ 161 | ("system", 162 | """You are a helpful assistant. You will have to answer to user's queries. 163 | You will have some context to help with your answers, but now always would be completely related or helpful. 164 | You can also use your knowledge to assist answering the user's queries.\n 165 | {context}"""), 166 | MessagesPlaceholder(variable_name="messages"), 167 | ("user", "{input}"), 168 | ]) 169 | stuff_documents_chain = create_stuff_documents_chain(llm, prompt) 170 | 171 | return create_retrieval_chain(retriever_chain, stuff_documents_chain) 172 | 173 | 174 | def stream_llm_rag_response(llm_stream, messages): 175 | if st.session_state.vector_db is None: 176 | st.error("Vector database is not initialized. Please upload some documents first.") 177 | return 178 | 179 | conversation_rag_chain = get_conversational_rag_chain(llm_stream) 180 | response_message = "*(RAG Response)*\n" 181 | try: 182 | for chunk in conversation_rag_chain.pick("answer").stream({"messages": messages[:-1], "input": messages[-1].content}): 183 | response_message += chunk 184 | yield chunk 185 | except Exception as e: 186 | st.error(f"An error occurred while processing the RAG response: {str(e)}") 187 | yield "I'm sorry, but I encountered an error while trying to process your request with the uploaded documents. Please try again or contact support if the issue persists." 188 | 189 | st.session_state.messages.append({"role": "assistant", "content": response_message}) 190 | -------------------------------------------------------------------------------- /notebooks/simple-product-info-chatbot.py: -------------------------------------------------------------------------------- 1 | import streamlit as st 2 | import os 3 | import dotenv 4 | import uuid 5 | from langchain.schema import HumanMessage, AIMessage 6 | from langchain_openai import ChatOpenAI 7 | from langchain_chroma import Chroma 8 | from langchain_openai import OpenAIEmbeddings 9 | 10 | from rag_methods import ( 11 | load_doc_to_db, 12 | load_url_to_db, 13 | stream_llm_response, 14 | stream_llm_rag_response, 15 | ) 16 | 17 | dotenv.load_dotenv() 18 | 19 | if "AZ_OPENAI_API_KEY" not in os.environ: 20 | MODELS = [ 21 | # "openai/o1-mini", 22 | "openai/gpt-4o", 23 | "openai/gpt-4o-mini", 24 | ] 25 | 26 | st.set_page_config( 27 | page_title="RAG LLM app?", 28 | page_icon="📚", 29 | layout="centered", 30 | initial_sidebar_state="expanded" 31 | ) 32 | 33 | 34 | # --- Header --- 35 | st.html("""

📚🔍 Do your LLM even RAG bro? 🤖💬

""") 36 | 37 | 38 | # --- Initial Setup --- 39 | if "session_id" not in st.session_state: 40 | st.session_state.session_id = str(uuid.uuid4()) 41 | 42 | if "rag_sources" not in st.session_state: 43 | st.session_state.rag_sources = [] 44 | 45 | if "messages" not in st.session_state: 46 | st.session_state.messages = [ 47 | {"role": "user", "content": "Hello"}, 48 | {"role": "assistant", "content": "Hi there! How can I assist you today?"} 49 | ] 50 | 51 | if "vector_db" not in st.session_state: 52 | st.session_state.vector_db = None 53 | 54 | persist_directory = f"./chroma_db_{st.session_state['session_id']}" 55 | if os.path.exists(persist_directory): 56 | embedding = OpenAIEmbeddings(api_key=st.session_state.openai_api_key) 57 | st.session_state.vector_db = Chroma(persist_directory=persist_directory, embedding_function=embedding) 58 | 59 | # --- Side Bar LLM API Tokens --- 60 | with st.sidebar: 61 | default_openai_api_key = os.getenv("OPENAI_API_KEY") if os.getenv("OPENAI_API_KEY") is not None else "" # only for development environment, otherwise it should return None 62 | with st.popover("🔐 OpenAI"): 63 | openai_api_key = st.text_input( 64 | "Introduce your OpenAI API Key (https://platform.openai.com/)", 65 | value=default_openai_api_key, 66 | type="password", 67 | key="openai_api_key", 68 | ) 69 | 70 | 71 | # --- Main Content --- 72 | # Checking if the user has introduced the OpenAI API Key, if not, a warning is displayed 73 | missing_openai = openai_api_key == "" or openai_api_key is None or "sk-" not in openai_api_key 74 | if missing_openai: 75 | st.write("#") 76 | st.warning("⬅️ Please introduce an API Key to continue...") 77 | 78 | else: 79 | # Sidebar 80 | with st.sidebar: 81 | st.divider() 82 | models = [] 83 | for model in MODELS: 84 | if "openai" in model and not missing_openai: 85 | models.append(model) 86 | 87 | st.selectbox( 88 | "🤖 Select a Model", 89 | options=models, 90 | key="model", 91 | ) 92 | 93 | cols0 = st.columns(2) 94 | with cols0[0]: 95 | is_vector_db_loaded = st.session_state.vector_db is not None 96 | st.toggle( 97 | "Use RAG", 98 | value=is_vector_db_loaded, 99 | key="use_rag", 100 | disabled=not is_vector_db_loaded, 101 | ) 102 | 103 | with cols0[1]: 104 | st.button("Clear Chat", on_click=lambda: st.session_state.messages.clear(), type="primary") 105 | 106 | st.header("RAG Sources:") 107 | 108 | # File upload input for RAG with documents 109 | st.file_uploader( 110 | "📄 Upload a document", 111 | type=["pdf", "txt", "docx", "md", "csv"], 112 | accept_multiple_files=True, 113 | on_change=load_doc_to_db, 114 | key="rag_docs", 115 | ) 116 | 117 | # URL input for RAG with websites 118 | st.text_input( 119 | "🌐 Introduce a URL", 120 | placeholder="https://example.com", 121 | on_change=load_url_to_db, 122 | key="rag_url", 123 | ) 124 | 125 | with st.expander(f"📚 Documents in DB ({0 if not is_vector_db_loaded else len(st.session_state.rag_sources)})"): 126 | st.write([] if not is_vector_db_loaded else [source for source in st.session_state.rag_sources]) 127 | 128 | 129 | # Main chat app 130 | model_provider = st.session_state.model.split("/")[0] 131 | if model_provider == "openai": 132 | llm_stream = ChatOpenAI( 133 | api_key=openai_api_key, 134 | model_name=st.session_state.model.split("/")[-1], 135 | temperature=0.3, 136 | streaming=True, 137 | ) 138 | 139 | for message in st.session_state.messages: 140 | with st.chat_message(message["role"]): 141 | st.markdown(message["content"]) 142 | 143 | if prompt := st.chat_input("Your message"): 144 | st.session_state.messages.append({"role": "user", "content": prompt}) 145 | with st.chat_message("user"): 146 | st.markdown(prompt) 147 | 148 | with st.chat_message("assistant"): 149 | message_placeholder = st.empty() 150 | full_response = "" 151 | 152 | messages = [HumanMessage(content=m["content"]) if m["role"] == "user" else AIMessage(content=m["content"]) for m in st.session_state.messages] 153 | 154 | if not st.session_state.use_rag or st.session_state.vector_db is None: 155 | st.write_stream(stream_llm_response(llm_stream, messages)) 156 | else: 157 | st.write_stream(stream_llm_rag_response(llm_stream, messages)) 158 | 159 | 160 | with st.sidebar: 161 | st.divider() 162 | st.video("https://youtu.be/abMwFViFFhI") 163 | st.write("📋[Medium Blog](https://medium.com/@enricdomingo/program-a-rag-llm-chat-app-with-langchain-streamlit-o1-gtp-4o-and-claude-3-5-529f0f164a5e)") 164 | st.write("📋[GitHub Repo](https://github.com/enricd/rag_llm_app)") 165 | 166 | 167 | -------------------------------------------------------------------------------- /notebooks/testing-langchain-app.ipynb: -------------------------------------------------------------------------------- 1 | { 2 | "cells": [ 3 | { 4 | "cell_type": "markdown", 5 | "metadata": {}, 6 | "source": [ 7 | "# Testing a LLM app\n", 8 | "\n", 9 | "1. Organize your parameters for experimentation\n", 10 | " 1. Atomic task that you're trying to solve\n", 11 | " 2. One or more prompts that solve either the whole thing or part of it\n", 12 | " 3. Define a metric that comprises performance\n", 13 | " 1. Summarization: g-eval method that involves consistency, relevancy, fluency, etc...\n", 14 | " 2. Extraction: produce a test coverage dataset with examples of examples of successful extraction and then work from there\n", 15 | " 3. General applications: LLM-as-a-judge just uses gpt-4o or gpt-4o-mini or whatever LLM to judge the outputs of other LLMs and score them for comparison and performance tracking\n", 16 | " 4. RAG: \n", 17 | " 1. Hallucination\n", 18 | " 2. Latency\n", 19 | " 3. Token\n", 20 | " 4. Token/cost \n", 21 | " 5. Benchmarks for your specific use case context/scenarion\n", 22 | " 6. Faithfulness (accuracy concerning the retrieved context, avoiding hallucinations), Context Relevancy (relevance of the retrieved context to the query), Answer Relevancy (relevance of the generated answer to the query)\n", 23 | " 7. Build your own small test dataset for whatever your task is, and iterate and grow it, by mixing manual annotations with semi-automated annotations and performance tracking\n", 24 | "2. Do a lot of few-shot examples in your prompts." 25 | ] 26 | } 27 | ], 28 | "metadata": { 29 | "language_info": { 30 | "name": "python" 31 | } 32 | }, 33 | "nbformat": 4, 34 | "nbformat_minor": 2 35 | } 36 | -------------------------------------------------------------------------------- /presentation_slides/2024-02-20-21-43-40.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/presentation_slides/2024-02-20-21-43-40.png -------------------------------------------------------------------------------- /presentation_slides/Getting-Started-with-LangChain-presentation.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/EnkrateiaLucca/oreilly_live_training_getting_started_with_langchain/9232c291a850aa62586ceadb8d7ebd453f2fa593/presentation_slides/Getting-Started-with-LangChain-presentation.pdf -------------------------------------------------------------------------------- /presentation_slides/extract-presentation.applescript: -------------------------------------------------------------------------------- 1 | tell application "Keynote" 2 | activate 3 | -- Make sure a presentation is open 4 | if not (exists front document) then error "Please open a presentation first." 5 | set thePresentation to front document 6 | set slideText to "" 7 | 8 | -- Loop through each slide in the presentation 9 | repeat with i from 1 to the count of slides of thePresentation 10 | set thisSlide to slide i of thePresentation 11 | set slideText to slideText & "Slide " & i & ":" 12 | 13 | -- Loop through each text item in the slide 14 | repeat with j from 1 to the count of text items of thisSlide 15 | set thisTextItem to text item j of thisSlide 16 | set theText to object text of thisTextItem 17 | set slideText to slideText & return & theText 18 | end repeat 19 | set slideText to slideText & return & return 20 | end repeat 21 | end tell 22 | 23 | -- Writing the extracted text to a file on the desktop 24 | set desktopPath to (path to desktop folder as text) & "KeynoteText.txt" 25 | set fileReference to open for access file desktopPath with write permission 26 | write slideText to fileReference 27 | close access fileReference 28 | 29 | -- Notifying the user that the script has finished 30 | display notification "Extracted text has been saved to KeynoteText.txt on your desktop." with title "Extraction Complete" -------------------------------------------------------------------------------- /presentation_slides/presentation-langgraph.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | Module 4: Getting Started with LangGraph 5 | 6 | 24 | 25 | 26 | 388 | 390 | 393 | 394 | -------------------------------------------------------------------------------- /presentation_slides/presentation.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | Introduction to LangChain 5 | 6 | 18 | 19 | 20 | 870 | 871 | 874 | 875 | 876 | 877 | ``` -------------------------------------------------------------------------------- /requirements/requirements.in: -------------------------------------------------------------------------------- 1 | ipykernel 2 | langchain>=0.3 3 | langchain-community 4 | langchain-openai 5 | langchain-ollama 6 | langchain_experimental 7 | langchain-cli 8 | langchain_chroma 9 | langchainhub 10 | langgraph>=0.2.58 11 | langsmith 12 | fastapi 13 | langserve 14 | openai 15 | pypdf 16 | faiss-cpu 17 | arxiv 18 | xmltodict 19 | duckduckgo-search 20 | playwright 21 | graphviz 22 | ipydrawio 23 | ipydrawio_widgets 24 | nbformat 25 | python-dotenv 26 | pandas 27 | matplotlib 28 | wikipedia 29 | google-search-results -------------------------------------------------------------------------------- /requirements/requirements.txt: -------------------------------------------------------------------------------- 1 | # This file was autogenerated by uv via the following command: 2 | # uv pip compile ./requirements/requirements.in -o ./requirements/requirements.txt 3 | aiofiles==22.1.0 4 | # via ypy-websocket 5 | aiohttp==3.9.5 6 | # via 7 | # langchain 8 | # langchain-community 9 | aiosignal==1.3.1 10 | # via aiohttp 11 | aiosqlite==0.20.0 12 | # via ypy-websocket 13 | annotated-types==0.7.0 14 | # via pydantic 15 | anyio==4.4.0 16 | # via 17 | # httpx 18 | # jupyter-server 19 | # openai 20 | # sse-starlette 21 | # starlette 22 | # watchfiles 23 | appnope==0.1.4 24 | # via ipykernel 25 | argon2-cffi==23.1.0 26 | # via 27 | # jupyter-server 28 | # notebook 29 | argon2-cffi-bindings==21.2.0 30 | # via argon2-cffi 31 | arrow==1.3.0 32 | # via isoduration 33 | arxiv==2.1.0 34 | # via -r ./requirements/requirements.in 35 | asgiref==3.8.1 36 | # via opentelemetry-instrumentation-asgi 37 | asttokens==2.4.1 38 | # via stack-data 39 | attrs==23.2.0 40 | # via 41 | # aiohttp 42 | # jsonschema 43 | # referencing 44 | babel==2.15.0 45 | # via jupyterlab-server 46 | backoff==2.2.1 47 | # via posthog 48 | bcrypt==4.1.3 49 | # via chromadb 50 | beautifulsoup4==4.12.3 51 | # via 52 | # nbconvert 53 | # wikipedia 54 | bleach==6.1.0 55 | # via nbconvert 56 | build==1.2.1 57 | # via chromadb 58 | cachetools==5.3.3 59 | # via google-auth 60 | certifi==2024.6.2 61 | # via 62 | # httpcore 63 | # httpx 64 | # kubernetes 65 | # requests 66 | cffi==1.16.0 67 | # via argon2-cffi-bindings 68 | charset-normalizer==3.3.2 69 | # via requests 70 | chroma-hnswlib==0.7.3 71 | # via chromadb 72 | chromadb==0.5.0 73 | # via langchain-chroma 74 | click==8.1.7 75 | # via 76 | # duckduckgo-search 77 | # typer 78 | # uvicorn 79 | colorama==0.4.6 80 | # via typer 81 | coloredlogs==15.0.1 82 | # via onnxruntime 83 | comm==0.2.2 84 | # via 85 | # ipykernel 86 | # ipywidgets 87 | contourpy==1.2.1 88 | # via matplotlib 89 | cycler==0.12.1 90 | # via matplotlib 91 | dataclasses-json==0.6.7 92 | # via langchain-community 93 | debugpy==1.8.1 94 | # via ipykernel 95 | decorator==5.1.1 96 | # via ipython 97 | defusedxml==0.7.1 98 | # via nbconvert 99 | deprecated==1.2.14 100 | # via 101 | # opentelemetry-api 102 | # opentelemetry-exporter-otlp-proto-grpc 103 | distro==1.9.0 104 | # via openai 105 | duckduckgo-search==6.1.6 106 | # via -r ./requirements/requirements.in 107 | entrypoints==0.4 108 | # via jupyter-client 109 | executing==2.0.1 110 | # via stack-data 111 | faiss-cpu==1.8.0 112 | # via -r ./requirements/requirements.in 113 | fastapi==0.110.3 114 | # via 115 | # -r ./requirements/requirements.in 116 | # chromadb 117 | # langchain-chroma 118 | # langserve 119 | # sse-starlette 120 | fastjsonschema==2.19.1 121 | # via nbformat 122 | feedparser==6.0.10 123 | # via arxiv 124 | filelock==3.15.1 125 | # via huggingface-hub 126 | flatbuffers==24.3.25 127 | # via onnxruntime 128 | fonttools==4.53.0 129 | # via matplotlib 130 | fqdn==1.5.1 131 | # via jsonschema 132 | frozenlist==1.4.1 133 | # via 134 | # aiohttp 135 | # aiosignal 136 | fsspec==2024.6.0 137 | # via huggingface-hub 138 | gitdb==4.0.11 139 | # via gitpython 140 | gitpython==3.1.43 141 | # via langchain-cli 142 | google-auth==2.30.0 143 | # via kubernetes 144 | google-search-results==2.4.2 145 | # via -r ./requirements/requirements.in 146 | googleapis-common-protos==1.63.1 147 | # via opentelemetry-exporter-otlp-proto-grpc 148 | graphviz==0.20.3 149 | # via -r ./requirements/requirements.in 150 | greenlet==3.0.3 151 | # via playwright 152 | grpcio==1.64.1 153 | # via 154 | # chromadb 155 | # opentelemetry-exporter-otlp-proto-grpc 156 | h11==0.14.0 157 | # via 158 | # httpcore 159 | # uvicorn 160 | httpcore==1.0.5 161 | # via httpx 162 | httptools==0.6.1 163 | # via uvicorn 164 | httpx==0.27.0 165 | # via 166 | # langgraph-sdk 167 | # langserve 168 | # langsmith 169 | # ollama 170 | # openai 171 | huggingface-hub==0.23.3 172 | # via tokenizers 173 | humanfriendly==10.0 174 | # via coloredlogs 175 | idna==3.7 176 | # via 177 | # anyio 178 | # httpx 179 | # jsonschema 180 | # requests 181 | # yarl 182 | importlib-metadata==7.1.0 183 | # via opentelemetry-api 184 | importlib-resources==6.4.0 185 | # via chromadb 186 | ipydrawio==1.3.0 187 | # via -r ./requirements/requirements.in 188 | ipydrawio-widgets==1.3.0 189 | # via 190 | # -r ./requirements/requirements.in 191 | # ipydrawio 192 | ipykernel==6.29.4 193 | # via 194 | # -r ./requirements/requirements.in 195 | # nbclassic 196 | # notebook 197 | ipython==8.25.0 198 | # via 199 | # ipykernel 200 | # ipywidgets 201 | # jupyterlab 202 | ipython-genutils==0.2.0 203 | # via 204 | # nbclassic 205 | # notebook 206 | ipywidgets==8.1.3 207 | # via ipydrawio-widgets 208 | isoduration==20.11.0 209 | # via jsonschema 210 | jedi==0.19.1 211 | # via ipython 212 | jinja2==3.1.4 213 | # via 214 | # jupyter-server 215 | # jupyterlab 216 | # jupyterlab-server 217 | # nbconvert 218 | # notebook 219 | jiter==0.6.1 220 | # via openai 221 | json5==0.9.25 222 | # via jupyterlab-server 223 | jsonpatch==1.33 224 | # via langchain-core 225 | jsonpointer==3.0.0 226 | # via 227 | # jsonpatch 228 | # jsonschema 229 | jsonschema==4.22.0 230 | # via 231 | # ipydrawio-widgets 232 | # jupyter-events 233 | # jupyterlab-server 234 | # nbformat 235 | jsonschema-specifications==2023.12.1 236 | # via jsonschema 237 | jupyter-client==7.4.9 238 | # via 239 | # ipykernel 240 | # jupyter-server 241 | # nbclient 242 | # notebook 243 | jupyter-core==5.7.2 244 | # via 245 | # ipykernel 246 | # jupyter-client 247 | # jupyter-server 248 | # jupyterlab 249 | # nbclient 250 | # nbconvert 251 | # nbformat 252 | # notebook 253 | jupyter-events==0.10.0 254 | # via 255 | # jupyter-server 256 | # jupyter-server-fileid 257 | jupyter-server==2.14.1 258 | # via 259 | # jupyter-server-fileid 260 | # jupyterlab 261 | # jupyterlab-server 262 | # notebook-shim 263 | jupyter-server-fileid==0.9.2 264 | # via jupyter-server-ydoc 265 | jupyter-server-terminals==0.5.3 266 | # via jupyter-server 267 | jupyter-server-ydoc==0.8.0 268 | # via jupyterlab 269 | jupyter-ydoc==0.2.5 270 | # via 271 | # jupyter-server-ydoc 272 | # jupyterlab 273 | jupyterlab==3.6.7 274 | # via ipydrawio 275 | jupyterlab-pygments==0.3.0 276 | # via nbconvert 277 | jupyterlab-server==2.27.2 278 | # via jupyterlab 279 | jupyterlab-widgets==3.0.11 280 | # via ipywidgets 281 | kiwisolver==1.4.5 282 | # via matplotlib 283 | kubernetes==30.1.0 284 | # via chromadb 285 | langchain==0.3.3 286 | # via 287 | # -r ./requirements/requirements.in 288 | # langchain-community 289 | langchain-chroma==0.1.4 290 | # via -r ./requirements/requirements.in 291 | langchain-cli==0.0.24 292 | # via -r ./requirements/requirements.in 293 | langchain-community==0.3.2 294 | # via 295 | # -r ./requirements/requirements.in 296 | # langchain-experimental 297 | langchain-core==0.3.24 298 | # via 299 | # langchain 300 | # langchain-chroma 301 | # langchain-community 302 | # langchain-experimental 303 | # langchain-ollama 304 | # langchain-openai 305 | # langchain-text-splitters 306 | # langgraph 307 | # langgraph-checkpoint 308 | # langserve 309 | langchain-experimental==0.3.2 310 | # via -r ./requirements/requirements.in 311 | langchain-ollama==0.2.0 312 | # via -r ./requirements/requirements.in 313 | langchain-openai==0.2.2 314 | # via -r ./requirements/requirements.in 315 | langchain-text-splitters==0.3.0 316 | # via langchain 317 | langchainhub==0.1.20 318 | # via -r ./requirements/requirements.in 319 | langgraph==0.2.58 320 | # via -r ./requirements/requirements.in 321 | langgraph-checkpoint==2.0.8 322 | # via langgraph 323 | langgraph-sdk==0.1.43 324 | # via langgraph 325 | langserve==0.3.0 326 | # via 327 | # -r ./requirements/requirements.in 328 | # langchain-cli 329 | langsmith==0.1.134 330 | # via 331 | # -r ./requirements/requirements.in 332 | # langchain 333 | # langchain-community 334 | # langchain-core 335 | libcst==1.4.0 336 | # via langchain-cli 337 | markdown-it-py==3.0.0 338 | # via rich 339 | markupsafe==2.1.5 340 | # via 341 | # jinja2 342 | # nbconvert 343 | marshmallow==3.21.3 344 | # via dataclasses-json 345 | matplotlib==3.9.0 346 | # via -r ./requirements/requirements.in 347 | matplotlib-inline==0.1.7 348 | # via 349 | # ipykernel 350 | # ipython 351 | mdurl==0.1.2 352 | # via markdown-it-py 353 | mistune==3.0.2 354 | # via nbconvert 355 | mmh3==4.1.0 356 | # via chromadb 357 | monotonic==1.6 358 | # via posthog 359 | mpmath==1.3.0 360 | # via sympy 361 | msgpack==1.1.0 362 | # via langgraph-checkpoint 363 | multidict==6.0.5 364 | # via 365 | # aiohttp 366 | # yarl 367 | mypy-extensions==1.0.0 368 | # via typing-inspect 369 | nbclassic==1.1.0 370 | # via 371 | # jupyterlab 372 | # notebook 373 | nbclient==0.10.0 374 | # via nbconvert 375 | nbconvert==7.16.4 376 | # via 377 | # jupyter-server 378 | # notebook 379 | nbformat==5.10.4 380 | # via 381 | # -r ./requirements/requirements.in 382 | # jupyter-server 383 | # nbclient 384 | # nbconvert 385 | # notebook 386 | nest-asyncio==1.6.0 387 | # via 388 | # ipykernel 389 | # jupyter-client 390 | # nbclassic 391 | # notebook 392 | notebook==6.5.7 393 | # via jupyterlab 394 | notebook-shim==0.2.4 395 | # via nbclassic 396 | numpy==1.26.4 397 | # via 398 | # chroma-hnswlib 399 | # chromadb 400 | # contourpy 401 | # faiss-cpu 402 | # langchain 403 | # langchain-chroma 404 | # langchain-community 405 | # matplotlib 406 | # onnxruntime 407 | # pandas 408 | oauthlib==3.2.2 409 | # via 410 | # kubernetes 411 | # requests-oauthlib 412 | ollama==0.3.1 413 | # via langchain-ollama 414 | onnxruntime==1.18.0 415 | # via chromadb 416 | openai==1.51.2 417 | # via 418 | # -r ./requirements/requirements.in 419 | # langchain-openai 420 | opentelemetry-api==1.25.0 421 | # via 422 | # chromadb 423 | # opentelemetry-exporter-otlp-proto-grpc 424 | # opentelemetry-instrumentation 425 | # opentelemetry-instrumentation-asgi 426 | # opentelemetry-instrumentation-fastapi 427 | # opentelemetry-sdk 428 | # opentelemetry-semantic-conventions 429 | opentelemetry-exporter-otlp-proto-common==1.25.0 430 | # via opentelemetry-exporter-otlp-proto-grpc 431 | opentelemetry-exporter-otlp-proto-grpc==1.25.0 432 | # via chromadb 433 | opentelemetry-instrumentation==0.46b0 434 | # via 435 | # opentelemetry-instrumentation-asgi 436 | # opentelemetry-instrumentation-fastapi 437 | opentelemetry-instrumentation-asgi==0.46b0 438 | # via opentelemetry-instrumentation-fastapi 439 | opentelemetry-instrumentation-fastapi==0.46b0 440 | # via chromadb 441 | opentelemetry-proto==1.25.0 442 | # via 443 | # opentelemetry-exporter-otlp-proto-common 444 | # opentelemetry-exporter-otlp-proto-grpc 445 | opentelemetry-sdk==1.25.0 446 | # via 447 | # chromadb 448 | # opentelemetry-exporter-otlp-proto-grpc 449 | opentelemetry-semantic-conventions==0.46b0 450 | # via 451 | # opentelemetry-instrumentation-asgi 452 | # opentelemetry-instrumentation-fastapi 453 | # opentelemetry-sdk 454 | opentelemetry-util-http==0.46b0 455 | # via 456 | # opentelemetry-instrumentation-asgi 457 | # opentelemetry-instrumentation-fastapi 458 | orjson==3.10.4 459 | # via 460 | # chromadb 461 | # duckduckgo-search 462 | # langgraph-sdk 463 | # langserve 464 | # langsmith 465 | overrides==7.7.0 466 | # via 467 | # chromadb 468 | # jupyter-server 469 | packaging==23.2 470 | # via 471 | # build 472 | # huggingface-hub 473 | # ipykernel 474 | # jupyter-server 475 | # jupyterlab 476 | # jupyterlab-server 477 | # langchain-core 478 | # langchainhub 479 | # marshmallow 480 | # matplotlib 481 | # nbconvert 482 | # onnxruntime 483 | pandas==2.2.2 484 | # via -r ./requirements/requirements.in 485 | pandocfilters==1.5.1 486 | # via nbconvert 487 | parso==0.8.4 488 | # via jedi 489 | pexpect==4.9.0 490 | # via ipython 491 | pillow==10.3.0 492 | # via matplotlib 493 | platformdirs==4.2.2 494 | # via jupyter-core 495 | playwright==1.44.0 496 | # via -r ./requirements/requirements.in 497 | posthog==3.5.0 498 | # via chromadb 499 | prometheus-client==0.20.0 500 | # via 501 | # jupyter-server 502 | # notebook 503 | prompt-toolkit==3.0.47 504 | # via ipython 505 | protobuf==4.25.3 506 | # via 507 | # googleapis-common-protos 508 | # onnxruntime 509 | # opentelemetry-proto 510 | psutil==5.9.8 511 | # via ipykernel 512 | ptyprocess==0.7.0 513 | # via 514 | # pexpect 515 | # terminado 516 | pure-eval==0.2.2 517 | # via stack-data 518 | pyasn1==0.6.0 519 | # via 520 | # pyasn1-modules 521 | # rsa 522 | pyasn1-modules==0.4.0 523 | # via google-auth 524 | pycparser==2.22 525 | # via cffi 526 | pydantic==2.7.4 527 | # via 528 | # chromadb 529 | # fastapi 530 | # langchain 531 | # langchain-core 532 | # langserve 533 | # langsmith 534 | # openai 535 | # pydantic-settings 536 | pydantic-core==2.18.4 537 | # via pydantic 538 | pydantic-settings==2.5.2 539 | # via langchain-community 540 | pyee==11.1.0 541 | # via playwright 542 | pygments==2.18.0 543 | # via 544 | # ipython 545 | # nbconvert 546 | # rich 547 | pyparsing==3.1.2 548 | # via matplotlib 549 | pypdf==4.2.0 550 | # via -r ./requirements/requirements.in 551 | pypika==0.48.9 552 | # via chromadb 553 | pyproject-hooks==1.1.0 554 | # via build 555 | pyreqwest-impersonate==0.4.7 556 | # via duckduckgo-search 557 | python-dateutil==2.9.0.post0 558 | # via 559 | # arrow 560 | # jupyter-client 561 | # kubernetes 562 | # matplotlib 563 | # pandas 564 | # posthog 565 | python-dotenv==1.0.1 566 | # via 567 | # -r ./requirements/requirements.in 568 | # pydantic-settings 569 | # uvicorn 570 | python-json-logger==2.0.7 571 | # via jupyter-events 572 | pytz==2024.1 573 | # via pandas 574 | pyyaml==6.0.1 575 | # via 576 | # chromadb 577 | # huggingface-hub 578 | # jupyter-events 579 | # kubernetes 580 | # langchain 581 | # langchain-community 582 | # langchain-core 583 | # libcst 584 | # uvicorn 585 | pyzmq==26.0.3 586 | # via 587 | # ipykernel 588 | # jupyter-client 589 | # jupyter-server 590 | # notebook 591 | referencing==0.35.1 592 | # via 593 | # jsonschema 594 | # jsonschema-specifications 595 | # jupyter-events 596 | regex==2024.5.15 597 | # via tiktoken 598 | requests==2.31.0 599 | # via 600 | # arxiv 601 | # chromadb 602 | # google-search-results 603 | # huggingface-hub 604 | # jupyterlab-server 605 | # kubernetes 606 | # langchain 607 | # langchain-community 608 | # langchainhub 609 | # langsmith 610 | # posthog 611 | # requests-oauthlib 612 | # requests-toolbelt 613 | # tiktoken 614 | # wikipedia 615 | requests-oauthlib==2.0.0 616 | # via kubernetes 617 | requests-toolbelt==1.0.0 618 | # via langsmith 619 | rfc3339-validator==0.1.4 620 | # via 621 | # jsonschema 622 | # jupyter-events 623 | rfc3986-validator==0.1.1 624 | # via 625 | # jsonschema 626 | # jupyter-events 627 | rich==13.7.1 628 | # via typer 629 | rpds-py==0.18.1 630 | # via 631 | # jsonschema 632 | # referencing 633 | rsa==4.9 634 | # via google-auth 635 | send2trash==1.8.3 636 | # via 637 | # jupyter-server 638 | # notebook 639 | setuptools==70.0.0 640 | # via opentelemetry-instrumentation 641 | sgmllib3k==1.0.0 642 | # via feedparser 643 | shellingham==1.5.4 644 | # via typer 645 | six==1.16.0 646 | # via 647 | # asttokens 648 | # bleach 649 | # kubernetes 650 | # posthog 651 | # python-dateutil 652 | # rfc3339-validator 653 | smmap==5.0.1 654 | # via gitdb 655 | sniffio==1.3.1 656 | # via 657 | # anyio 658 | # httpx 659 | # openai 660 | soupsieve==2.5 661 | # via beautifulsoup4 662 | sqlalchemy==2.0.30 663 | # via 664 | # langchain 665 | # langchain-community 666 | sse-starlette==1.8.2 667 | # via langserve 668 | stack-data==0.6.3 669 | # via ipython 670 | starlette==0.37.2 671 | # via 672 | # fastapi 673 | # sse-starlette 674 | sympy==1.12.1 675 | # via onnxruntime 676 | tenacity==8.3.0 677 | # via 678 | # chromadb 679 | # langchain 680 | # langchain-community 681 | # langchain-core 682 | terminado==0.18.1 683 | # via 684 | # jupyter-server 685 | # jupyter-server-terminals 686 | # notebook 687 | tiktoken==0.7.0 688 | # via langchain-openai 689 | tinycss2==1.3.0 690 | # via nbconvert 691 | tokenizers==0.19.1 692 | # via chromadb 693 | tomlkit==0.12.5 694 | # via langchain-cli 695 | tornado==6.4.1 696 | # via 697 | # ipykernel 698 | # jupyter-client 699 | # jupyter-server 700 | # jupyterlab 701 | # notebook 702 | # terminado 703 | tqdm==4.66.4 704 | # via 705 | # chromadb 706 | # huggingface-hub 707 | # openai 708 | traitlets==5.14.3 709 | # via 710 | # comm 711 | # ipykernel 712 | # ipython 713 | # ipywidgets 714 | # jupyter-client 715 | # jupyter-core 716 | # jupyter-events 717 | # jupyter-server 718 | # matplotlib-inline 719 | # nbclient 720 | # nbconvert 721 | # nbformat 722 | # notebook 723 | typer==0.9.4 724 | # via 725 | # chromadb 726 | # langchain-cli 727 | types-python-dateutil==2.9.0.20240316 728 | # via arrow 729 | types-requests==2.32.0.20240602 730 | # via langchainhub 731 | typing-extensions==4.12.2 732 | # via 733 | # aiosqlite 734 | # chromadb 735 | # fastapi 736 | # huggingface-hub 737 | # ipython 738 | # langchain-core 739 | # openai 740 | # opentelemetry-sdk 741 | # pydantic 742 | # pydantic-core 743 | # pyee 744 | # sqlalchemy 745 | # typer 746 | # typing-inspect 747 | typing-inspect==0.9.0 748 | # via dataclasses-json 749 | tzdata==2024.1 750 | # via pandas 751 | uri-template==1.3.0 752 | # via jsonschema 753 | urllib3==2.2.1 754 | # via 755 | # kubernetes 756 | # requests 757 | # types-requests 758 | uvicorn==0.23.2 759 | # via 760 | # chromadb 761 | # langchain-cli 762 | # sse-starlette 763 | uvloop==0.19.0 764 | # via uvicorn 765 | watchfiles==0.22.0 766 | # via uvicorn 767 | wcwidth==0.2.13 768 | # via prompt-toolkit 769 | webcolors==24.6.0 770 | # via jsonschema 771 | webencodings==0.5.1 772 | # via 773 | # bleach 774 | # tinycss2 775 | websocket-client==1.8.0 776 | # via 777 | # jupyter-server 778 | # kubernetes 779 | websockets==12.0 780 | # via uvicorn 781 | widgetsnbextension==4.0.11 782 | # via ipywidgets 783 | wikipedia==1.4.0 784 | # via -r ./requirements/requirements.in 785 | wrapt==1.16.0 786 | # via 787 | # deprecated 788 | # opentelemetry-instrumentation 789 | xmltodict==0.13.0 790 | # via -r ./requirements/requirements.in 791 | y-py==0.6.2 792 | # via 793 | # jupyter-ydoc 794 | # ypy-websocket 795 | yarl==1.9.4 796 | # via aiohttp 797 | ypy-websocket==0.8.4 798 | # via jupyter-server-ydoc 799 | zipp==3.19.2 800 | # via importlib-metadata 801 | --------------------------------------------------------------------------------