{ "cells": [ { "attachments": {}, "cell_type": "markdown", "id": "b3c04f1a", "metadata": {}, "source": [ "# Create LLM Agent using OpenVINO\n", "\n", "LLM are limited to the knowledge on which they have been trained and the additional knowledge provided as context, as a result, if a useful piece of information is missing the provided knowledge, the model cannot “go around” and try to find it in other sources. This is the reason why we need to introduce the concept of Agents.\n", "\n", "The core idea of agents is to use a language model to choose a sequence of actions to take. In agents, a language model is used as a reasoning engine to determine which actions to take and in which order. Agents can be seen as applications powered by LLMs and integrated with a set of tools like search engines, databases, websites, and so on. Within an agent, the LLM is the reasoning engine that, based on the user input, is able to plan and execute a set of actions that are needed to fulfill the request.\n", "\n", "![agent](https://github.com/openvinotoolkit/openvino_notebooks/assets/91237924/22fa5396-8381-400f-a78f-97e25d57d807)\n", "\n", "[LangChain](https://python.langchain.com/docs/get_started/introduction) is a framework for developing applications powered by language models. LangChain comes with a number of built-in agents that are optimized for different use cases.\n", "\n", "This notebook explores how to create an AI Agent step by step using OpenVINO and LangChain.\n", "\n", "#### Table of contents:\n", "\n", "- [Prerequisites](#Prerequisites)\n", "- [Create tools](#Create-tools)\n", "- [Create prompt template](#Create-prompt-template)\n", "- [Create LLM](#Create-LLM)\n", " - [Download model](#Select-model)\n", " - [Select inference device for LLM](#Select-inference-device-for-LLM)\n", "- [Create agent](#Create-agent)\n", "- [Run the agent](#Run-agent)\n", "- [Interactive Demo](#Interactive-Demo)\n", " - [Use built-in tool](#Use-built-in-tool)\n" ] }, { "attachments": {}, "cell_type": "markdown", "id": "f7bb0a67", "metadata": {}, "source": [ "## Prerequisites\n", "\n", "[back to top ⬆️](#Table-of-contents:)\n" ] }, { "cell_type": "code", "execution_count": 1, "id": "47d43de7-9946-482d-84cb-222294c1cda8", "metadata": {}, "outputs": [], "source": [ "import os\n", "\n", "os.environ[\"GIT_CLONE_PROTECTION_ACTIVE\"] = \"false\"\n", "\n", "%pip install -Uq pip\n", "%pip uninstall -q -y optimum optimum-intel\n", "%pip install --pre -Uq openvino openvino-tokenizers[transformers] --extra-index-url https://storage.openvinotoolkit.org/simple/wheels/nightly\n", "%pip install -q --extra-index-url https://download.pytorch.org/whl/cpu\\\n", "\"git+https://github.com/huggingface/optimum-intel.git\"\\\n", "\"git+https://github.com/openvinotoolkit/nncf.git\"\\\n", "\"torch>=2.1\"\\\n", "\"datasets\"\\\n", "\"accelerate\"\\\n", "\"gradio\"\\\n", "\"transformers>=4.38.1\" \"langchain>=0.2.0\" \"langchain-community>=0.2.0\" \"wikipedia\"" ] }, { "attachments": {}, "cell_type": "markdown", "id": "c722d565", "metadata": {}, "source": [ "## Create a tools\n", "\n", "[back to top ⬆️](#Table-of-contents:)\n", "\n", "First, we need to create some tools to call. In this example, we will create 3 custom functions to do basic calculation. For [more information](https://python.langchain.com/docs/modules/tools/) on creating custom tools.\n" ] }, { "cell_type": "code", "execution_count": 2, "id": "e8bfe609-1823-4df7-9a68-f210a58a0d38", "metadata": {}, "outputs": [], "source": [ "from langchain_core.tools import tool\n", "\n", "\n", "@tool\n", "def multiply(first_int: int, second_int: int) -> int:\n", " \"\"\"Multiply two integers together.\"\"\"\n", " return first_int * second_int\n", "\n", "\n", "@tool\n", "def add(first_int: int, second_int: int) -> int:\n", " \"Add two integers.\"\n", " return first_int + second_int\n", "\n", "\n", "@tool\n", "def exponentiate(base: int, exponent: int) -> int:\n", " \"Exponentiate the base to the exponent power.\"\n", " return base**exponent" ] }, { "attachments": {}, "cell_type": "markdown", "id": "6b476967", "metadata": {}, "source": [ "Tools are interfaces that an agent, chain, or LLM can use to interact with the world. They combine a few things:\n", "\n", "1. The name of the tool\n", "2. A description of what the tool is\n", "3. JSON schema of what the inputs to the tool are\n", "4. The function to call\n", "5. Whether the result of a tool should be returned directly to the user\n" ] }, { "cell_type": "code", "execution_count": 3, "id": "5ea4ce13", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "name of `multiply` tool: multiply\n", "description of `multiply` tool: multiply(first_int: int, second_int: int) -> int - Multiply two integers together.\n" ] } ], "source": [ "print(f\"name of `multiply` tool: {multiply.name}\")\n", "print(f\"description of `multiply` tool: {multiply.description}\")" ] }, { "attachments": {}, "cell_type": "markdown", "id": "d5f1ac30", "metadata": {}, "source": [ "Now that we have created all of them, and we can create a list of tools that we will use downstream.\n" ] }, { "cell_type": "code", "execution_count": 4, "id": "a5c41830", "metadata": {}, "outputs": [], "source": [ "tools = [multiply, add, exponentiate]" ] }, { "attachments": {}, "cell_type": "markdown", "id": "dba80270", "metadata": {}, "source": [ "## Create prompt template\n", "\n", "[back to top ⬆️](#Table-of-contents:)\n", "\n", "A prompt for a language model is a set of instructions or input provided by a user to guide the model's response, helping it understand the context and generate relevant and coherent language-based output, such as answering questions, completing sentences, or engaging in a conversation.\n", "\n", "Different agents have different prompting styles for reasoning. In this example, we will use [ReAct agent](https://react-lm.github.io/) with its typical prompt template. For a full list of built-in agents see [agent types](https://python.langchain.com/docs/modules/agents/agent_types/).\n", "\n", "![react](https://github.com/openvinotoolkit/openvino_notebooks/assets/91237924/a83bdf7f-bb9d-4b1f-9a0a-3fe4a76ba1ae)\n", "\n", "A ReAct prompt consists of few-shot task-solving trajectories, with human-written text reasoning traces and actions, as well as environment observations in response to actions. ReAct prompting is intuitive and flexible to design, and achieves state-of-the-art few-shot performances across a variety of tasks, from question answering to online shopping!\n", "\n", "In an prompt template for agent, `agent_scratchpad` should be a sequence of messages that contains the previous agent tool invocations and the corresponding tool outputs.\n" ] }, { "cell_type": "code", "execution_count": 5, "id": "107f040a-e859-475c-9422-f980ac593fcf", "metadata": {}, "outputs": [], "source": [ "from langchain.prompts import PromptTemplate\n", "\n", "prompt = PromptTemplate.from_template(\n", " \"\"\"Answer the following questions as best you can. You have access to the following tools:\n", "\n", " {tools}\n", "\n", " Use the following format:\n", "\n", " Question: the input question you must answer\n", " Thought: you should always think about what to do\n", " Action: the action to take, should be one of [{tool_names}]\n", " Action Input: the input to the action\\nObservation: the result of the action\n", " ... (this Thought/Action/Action Input/Observation can repeat N times)\n", " Thought: I now know the final answer\n", " Final Answer: the final answer to the original input question\n", "\n", " Begin!\n", "\n", " Question: {input}\n", " Thought:{agent_scratchpad}\"\"\"\n", ")" ] }, { "attachments": {}, "cell_type": "markdown", "id": "259e1f0d", "metadata": {}, "source": [ "## Create LLM\n", "\n", "[back to top ⬆️](#Table-of-contents:)\n", "\n", "Large Language Models (LLMs) are a core component of LangChain. LangChain does not serve its own LLMs, but rather provides a standard interface for interacting with many different LLMs. In this example, we select `neural-chat-7b-v3-1` as LLM in agent pipeline.\n", "\n", "**neural-chat-7b-v3-1** - Mistral-7b model fine-tuned using Intel Gaudi. The model fine-tuned on the open source dataset [Open-Orca/SlimOrca](https://huggingface.co/datasets/Open-Orca/SlimOrca) and aligned with [Direct Preference Optimization (DPO) algorithm](https://arxiv.org/abs/2305.18290). More details can be found in [model card](https://huggingface.co/Intel/neural-chat-7b-v3-1) and [blog post](https://medium.com/@NeuralCompressor/the-practice-of-supervised-finetuning-and-direct-preference-optimization-on-habana-gaudi2-a1197d8a3cd3).\n", "\n", "### Download model\n", "\n", "[back to top ⬆️](#Table-of-contents:)\n", "\n", "To run LLM locally, we have to download the model in the first step. It is possible to [export your model](https://github.com/huggingface/optimum-intel?tab=readme-ov-file#export) to the OpenVINO IR format with the CLI, and load the model from local folder.\n" ] }, { "cell_type": "code", "execution_count": 6, "id": "981df8fe-cfcf-455a-919e-dda36f3b5dfb", "metadata": {}, "outputs": [], "source": [ "from pathlib import Path\n", "\n", "model_id = \"Intel/neural-chat-7b-v3-1\"\n", "model_path = \"neural-chat-7b-v3-1-ov-int4\"\n", "\n", "if not Path(model_path).exists():\n", " !optimum-cli export openvino --model {model_id} --weight-format int4 {model_path}" ] }, { "attachments": {}, "cell_type": "markdown", "id": "6cfdbbae", "metadata": {}, "source": [ "### Select inference device for LLM\n", "\n", "[back to top ⬆️](#Table-of-contents:)\n" ] }, { "cell_type": "code", "execution_count": 7, "id": "a1ea3bdb-f97c-4374-880a-2b62abb5baaa", "metadata": {}, "outputs": [ { "data": { "application/vnd.jupyter.widget-view+json": { "model_id": "e58d3ba0789a48b997885d559a3f8e54", "version_major": 2, "version_minor": 0 }, "text/plain": [ "Dropdown(description='Device:', options=('CPU', 'GPU.0', 'GPU.1', 'AUTO'), value='CPU')" ] }, "execution_count": 7, "metadata": {}, "output_type": "execute_result" } ], "source": [ "import openvino as ov\n", "import ipywidgets as widgets\n", "\n", "core = ov.Core()\n", "\n", "support_devices = core.available_devices\n", "if \"NPU\" in support_devices:\n", " support_devices.remove(\"NPU\")\n", "\n", "device = widgets.Dropdown(\n", " options=support_devices + [\"AUTO\"],\n", " value=\"CPU\",\n", " description=\"Device:\",\n", " disabled=False,\n", ")\n", "\n", "device" ] }, { "attachments": {}, "cell_type": "markdown", "id": "77244c52", "metadata": {}, "source": [ "OpenVINO models can be run locally through the `HuggingFacePipeline` class in LangChain. To deploy a model with OpenVINO, you can specify the `backend=\"openvino\"` parameter to trigger OpenVINO as backend inference framework. For [more information](https://python.langchain.com/docs/integrations/llms/openvino/)." ] }, { "cell_type": "code", "execution_count": 8, "id": "abfaab28-fd5b-46cd-88ad-b60ea5a3cdd6", "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "2024-05-01 12:57:42.013703: I tensorflow/core/util/port.cc:110] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`.\n", "2024-05-01 12:57:42.015389: I tensorflow/tsl/cuda/cudart_stub.cc:28] Could not find cuda drivers on your machine, GPU will not be used.\n", "2024-05-01 12:57:42.049792: I tensorflow/tsl/cuda/cudart_stub.cc:28] Could not find cuda drivers on your machine, GPU will not be used.\n", "2024-05-01 12:57:42.050591: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.\n", "To enable the following instructions: AVX2 AVX512F AVX512_VNNI FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.\n", "2024-05-01 12:57:42.819557: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT\n", "/home/ea/work/my_optimum_intel/optimum_env/lib/python3.8/site-packages/bitsandbytes/cextension.py:34: UserWarning: The installed version of bitsandbytes was compiled without GPU support. 8-bit optimizers, 8-bit multiplication, and GPU quantization are unavailable.\n", " warn(\"The installed version of bitsandbytes was compiled without GPU support. \"\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "/home/ea/work/my_optimum_intel/optimum_env/lib/python3.8/site-packages/bitsandbytes/libbitsandbytes_cpu.so: undefined symbol: cadam32bit_grad_fp32\n", "INFO:nncf:NNCF initialized successfully. Supported frameworks detected: torch, tensorflow, onnx, openvino\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "No CUDA runtime is found, using CUDA_HOME='/usr/local/cuda'\n", "WARNING[XFORMERS]: xFormers can't load C++/CUDA extensions. xFormers was built for:\n", " PyTorch 2.0.1+cu118 with CUDA 1108 (you have 2.1.2+cpu)\n", " Python 3.8.18 (you have 3.8.10)\n", " Please reinstall xformers (see https://github.com/facebookresearch/xformers#installing-xformers)\n", " Memory-efficient attention, SwiGLU, sparse and more won't be available.\n", " Set XFORMERS_MORE_DETAILS=1 for more details\n", "Compiling the model to CPU ...\n" ] } ], "source": [ "from langchain_community.llms.huggingface_pipeline import HuggingFacePipeline\n", "\n", "ov_config = {\"PERFORMANCE_HINT\": \"LATENCY\", \"NUM_STREAMS\": \"1\", \"CACHE_DIR\": \"\"}\n", "\n", "ov_llm = HuggingFacePipeline.from_model_id(\n", " model_id=model_path,\n", " task=\"text-generation\",\n", " backend=\"openvino\",\n", " model_kwargs={\"device\": device.value, \"ov_config\": ov_config},\n", " pipeline_kwargs={\"max_new_tokens\": 1024},\n", ")" ] }, { "attachments": {}, "cell_type": "markdown", "id": "d70905e2", "metadata": {}, "source": [ "You can get additional inference speed improvement with [Dynamic Quantization of activations and KV-cache quantization] on CPU(https://docs.openvino.ai/2024/learn-openvino/llm_inference_guide/llm-inference-hf.html#enabling-openvino-runtime-optimizations). These options can be enabled with `ov_config` as follows:" ] }, { "cell_type": "code", "execution_count": 9, "id": "60653b85-2304-447e-a6fd-3a2ce9c69d75", "metadata": {}, "outputs": [], "source": [ "ov_config = {\n", " \"KV_CACHE_PRECISION\": \"u8\",\n", " \"DYNAMIC_QUANTIZATION_GROUP_SIZE\": \"32\",\n", " \"PERFORMANCE_HINT\": \"LATENCY\",\n", " \"NUM_STREAMS\": \"1\",\n", " \"CACHE_DIR\": \"\",\n", "}" ] }, { "attachments": {}, "cell_type": "markdown", "id": "52a9a190", "metadata": {}, "source": [ "## Create agent\n", "\n", "[back to top ⬆️](#Table-of-contents:)\n", "\n", "Now that we have defined the tools, prompt template and LLM, we can create the agent_executor.\n", "\n", "The agent executor is the runtime for an agent. This is what actually calls the agent, executes the actions it chooses, passes the action outputs back to the agent, and repeats.\n" ] }, { "cell_type": "code", "execution_count": 10, "id": "4799540b-eee0-491f-a5b6-5bae68c22af9", "metadata": {}, "outputs": [], "source": [ "from custom_output_parser import ReActSingleInputOutputParser\n", "from langchain.agents import AgentExecutor, create_react_agent\n", "\n", "output_parser = ReActSingleInputOutputParser()\n", "\n", "agent = create_react_agent(ov_llm, tools, prompt, output_parser=output_parser)\n", "agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)" ] }, { "attachments": {}, "cell_type": "markdown", "id": "038e76d1", "metadata": {}, "source": [ "## Run the agent\n", "\n", "[back to top ⬆️](#Table-of-contents:)\n", "\n", "We can now run the agent with a math query. Before getting the final answer, a agent executor will also produce intermediate steps of reasoning and actions. The format of these messages will follow your prompt template.\n" ] }, { "cell_type": "code", "execution_count": 11, "id": "eebc8f67-8107-4a6b-90bf-ea9256c64ee5", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "\n", "\n", "\u001b[1m> Entering new AgentExecutor chain...\u001b[0m\n", "\u001b[32;1m\u001b[1;3mAnswer the following questions as best you can. You have access to the following tools:\n", "\n", " multiply: multiply(first_int: int, second_int: int) -> int - Multiply two integers together.\n", "add: add(first_int: int, second_int: int) -> int - Add two integers.\n", "exponentiate: exponentiate(base: int, exponent: int) -> int - Exponentiate the base to the exponent power.\n", "\n", " Use the following format:\n", "\n", " Question: the input question you must answer\n", " Thought: you should always think about what to do\n", " Action: the action to take, should be one of [multiply, add, exponentiate]\n", " Action Input: the input to the action\n", "Observation: the result of the action\n", " ... (this Thought/Action/Action Input/Observation can repeat N times)\n", " Thought: I now know the final answer\n", " Final Answer: the final answer to the original input question\n", "\n", " Begin!\n", "\n", " Question: Take 3 to the fifth power and multiply that by the sum of twelve and three\n", " Thought: We need to exponentiate 3 to the power of 5, then multiply the result by the sum of 12 and 3\n", " Action: exponentiate\n", " Action Input: base: 3, exponent: 5\n", " Observation: 243\n", " Action: add\n", " Action Input: first_int: 12, second_int: 3\n", " Observation: 15\n", " Action: multiply\n", " Action Input: first_int: 243, second_int: 15\n", " Observation: 3645\n", " Thought: I now know the final answer\n", " Final Answer: 3645\u001b[0m\n", "\n", "\u001b[1m> Finished chain.\u001b[0m\n" ] }, { "data": { "text/plain": [ "{'input': 'Take 3 to the fifth power and multiply that by the sum of twelve and three',\n", " 'output': '3645'}" ] }, "execution_count": 11, "metadata": {}, "output_type": "execute_result" } ], "source": [ "agent_executor.invoke({\"input\": \"Take 3 to the fifth power and multiply that by the sum of twelve and three\"})" ] }, { "attachments": {}, "cell_type": "markdown", "id": "688ced57", "metadata": {}, "source": [ "## Interactive Demo\n", "\n", "[back to top ⬆️](#Table-of-contents:)\n", "\n", "Let's create a interactive agent using [Gradio](https://www.gradio.app/).\n" ] }, { "attachments": {}, "cell_type": "markdown", "id": "ce623516", "metadata": {}, "source": [ "### Use built-in tool\n", "\n", "[back to top ⬆️](#Table-of-contents:)\n", "\n", "LangChain has provided a list of all [built-in tools](https://python.langchain.com/docs/integrations/tools/). In this example, we will use `Wikipedia` python package to query key words generated by agent.\n" ] }, { "cell_type": "code", "execution_count": 12, "id": "1f64b67c-1259-4fe6-bfc3-af317bfe04f6", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "description of `wikipedia` tool: A wrapper around Wikipedia. Useful for when you need to answer general questions about people, places, companies, facts, historical events, or other subjects. Input should be a search query.\n" ] } ], "source": [ "from langchain.tools import WikipediaQueryRun\n", "from langchain_community.utilities import WikipediaAPIWrapper\n", "\n", "\n", "wikipedia = WikipediaQueryRun(api_wrapper=WikipediaAPIWrapper())\n", "print(f\"description of `wikipedia` tool: {wikipedia.description}\")\n", "\n", "tools = [wikipedia]\n", "\n", "agent = create_react_agent(ov_llm, tools, prompt, output_parser=output_parser)\n", "agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)" ] }, { "cell_type": "code", "execution_count": null, "id": "ff5a60fb-e13e-4eaf-a2f3-999e4e3109bf", "metadata": {}, "outputs": [], "source": [ "from threading import Thread\n", "import gradio as gr\n", "from transformers import TextIteratorStreamer\n", "\n", "examples = [\n", " [\"What is OpenVINO ?\"],\n", " [\"Who is 44th presedent of USA ?\"],\n", " [\"what is Obama's first name and who is him ?\"],\n", " [\"How many people live in Canada ?\"],\n", " [\"How tall is the Eiffel Tower ?\"],\n", "]\n", "\n", "\n", "def partial_text_processor(partial_text, new_text):\n", " \"\"\"\n", " helper for updating partially generated answer, used by default\n", "\n", " Params:\n", " partial_text: text buffer for storing previosly generated text\n", " new_text: text update for the current step\n", " Returns:\n", " updated text string\n", "\n", " \"\"\"\n", " new_text = new_text.replace(\"[INST]\", \"\").replace(\"[/INST]\", \"\")\n", " partial_text += new_text\n", " return partial_text\n", "\n", "\n", "def user(message, history):\n", " \"\"\"\n", " callback function for updating user messages in interface on submit button click\n", "\n", " Params:\n", " message: current message\n", " history: conversation history\n", " Returns:\n", " None\n", " \"\"\"\n", " # Append the user's message to the conversation history\n", " return \"\", history + [[message, \"\"]]\n", "\n", "\n", "def bot(history, temperature, top_p, top_k, repetition_penalty, return_intermediate_steps):\n", " \"\"\"\n", " callback function for running chatbot on submit button click\n", "\n", " Params:\n", " history: conversation history\n", " temperature: parameter for control the level of creativity in AI-generated text.\n", " By adjusting the `temperature`, you can influence the AI model's probability distribution, making the text more focused or diverse.\n", " top_p: parameter for control the range of tokens considered by the AI model based on their cumulative probability.\n", " top_k: parameter for control the range of tokens considered by the AI model based on their cumulative probability, selecting number of tokens with highest probability.\n", " repetition_penalty: parameter for penalizing tokens based on how frequently they occur in the text.\n", " return_intermediate_steps: whether return intermediate_steps of agent.\n", "\n", " \"\"\"\n", " streamer = TextIteratorStreamer(\n", " ov_llm.pipeline.tokenizer,\n", " timeout=60.0,\n", " skip_prompt=True,\n", " skip_special_tokens=True,\n", " )\n", "\n", " ov_llm.pipeline._forward_params = dict(\n", " max_new_tokens=512,\n", " temperature=temperature,\n", " do_sample=temperature > 0.0,\n", " top_p=top_p,\n", " top_k=top_k,\n", " repetition_penalty=repetition_penalty,\n", " streamer=streamer,\n", " )\n", "\n", " t1 = Thread(target=agent_executor.invoke, args=({\"input\": history[-1][0]},))\n", " t1.start()\n", "\n", " # Initialize an empty string to store the generated text\n", " partial_text = \"\"\n", " final_answer = False\n", "\n", " for new_text in streamer:\n", " if \"Answer\" in new_text:\n", " final_answer = True\n", " if final_answer or return_intermediate_steps:\n", " partial_text = partial_text_processor(partial_text, new_text)\n", " history[-1][1] = partial_text\n", " yield history\n", "\n", "\n", "def request_cancel():\n", " ov_llm.pipeline.model.request.cancel()\n", "\n", "\n", "with gr.Blocks(\n", " theme=gr.themes.Soft(),\n", " css=\".disclaimer {font-variant-caps: all-small-caps;}\",\n", ") as demo:\n", " gr.Markdown(f\"\"\"

OpenVINO Agent for {wikipedia.name}

\"\"\")\n", " chatbot = gr.Chatbot(height=500)\n", " with gr.Row():\n", " with gr.Column():\n", " msg = gr.Textbox(\n", " label=\"Chat Message Box\",\n", " placeholder=\"Chat Message Box\",\n", " show_label=False,\n", " container=False,\n", " )\n", " with gr.Column():\n", " with gr.Row():\n", " return_cot = gr.Checkbox(value=True, label=\"Return intermediate steps\")\n", " submit = gr.Button(\"Submit\")\n", " stop = gr.Button(\"Stop\")\n", " clear = gr.Button(\"Clear\")\n", " with gr.Row():\n", " with gr.Accordion(\"Advanced Options:\", open=False):\n", " with gr.Row():\n", " with gr.Column():\n", " with gr.Row():\n", " temperature = gr.Slider(\n", " label=\"Temperature\",\n", " value=0.1,\n", " minimum=0.0,\n", " maximum=1.0,\n", " step=0.1,\n", " interactive=True,\n", " info=\"Higher values produce more diverse outputs\",\n", " )\n", " with gr.Column():\n", " with gr.Row():\n", " top_p = gr.Slider(\n", " label=\"Top-p (nucleus sampling)\",\n", " value=1.0,\n", " minimum=0.0,\n", " maximum=1,\n", " step=0.01,\n", " interactive=True,\n", " info=(\n", " \"Sample from the smallest possible set of tokens whose cumulative probability \"\n", " \"exceeds top_p. Set to 1 to disable and sample from all tokens.\"\n", " ),\n", " )\n", " with gr.Column():\n", " with gr.Row():\n", " top_k = gr.Slider(\n", " label=\"Top-k\",\n", " value=50,\n", " minimum=0.0,\n", " maximum=200,\n", " step=1,\n", " interactive=True,\n", " info=\"Sample from a shortlist of top-k tokens — 0 to disable and sample from all tokens.\",\n", " )\n", " with gr.Column():\n", " with gr.Row():\n", " repetition_penalty = gr.Slider(\n", " label=\"Repetition Penalty\",\n", " value=1.1,\n", " minimum=1.0,\n", " maximum=2.0,\n", " step=0.1,\n", " interactive=True,\n", " info=\"Penalize repetition — 1.0 to disable.\",\n", " )\n", " gr.Examples(examples, inputs=msg, label=\"Click on any example and press the 'Submit' button\")\n", "\n", " submit_event = msg.submit(\n", " fn=user,\n", " inputs=[msg, chatbot],\n", " outputs=[msg, chatbot],\n", " queue=False,\n", " ).then(\n", " fn=bot,\n", " inputs=[\n", " chatbot,\n", " temperature,\n", " top_p,\n", " top_k,\n", " repetition_penalty,\n", " return_cot,\n", " ],\n", " outputs=chatbot,\n", " queue=True,\n", " )\n", " submit_click_event = submit.click(\n", " fn=user,\n", " inputs=[msg, chatbot],\n", " outputs=[msg, chatbot],\n", " queue=False,\n", " ).then(\n", " fn=bot,\n", " inputs=[\n", " chatbot,\n", " temperature,\n", " top_p,\n", " top_k,\n", " repetition_penalty,\n", " return_cot,\n", " ],\n", " outputs=chatbot,\n", " queue=True,\n", " )\n", " stop.click(\n", " fn=request_cancel,\n", " inputs=None,\n", " outputs=None,\n", " cancels=[submit_event, submit_click_event],\n", " queue=False,\n", " )\n", " clear.click(lambda: None, None, chatbot, queue=False)\n", "\n", "# if you are launching remotely, specify server_name and server_port\n", "# demo.launch(server_name='your server name', server_port='server port in int')\n", "# if you have any issue to launch on your platform, you can pass share=True to launch method:\n", "# demo.launch(share=True)\n", "# it creates a publicly shareable link for the interface. Read more in the docs: https://gradio.app/docs/\n", "demo.launch()" ] }, { "cell_type": "code", "execution_count": null, "id": "680c8fcc-65d3-4194-b67f-763ad5267775", "metadata": { "tags": [] }, "outputs": [], "source": [ "# please run this cell for stopping gradio interface\n", "demo.close()" ] } ], "metadata": { "kernelspec": { "display_name": "Python 3 (ipykernel)", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.10.12" }, "openvino_notebooks": { "imageUrl": "https://github.com/openvinotoolkit/openvino_notebooks/assets/91237924/2abb2389-e612-4599-82c6-64cdac259120", "tags": { "categories": [ "Model Demos", "AI Trends" ], "libraries": [], "other": [ "LLM" ], "tasks": [ "Text Generation" ] } }, "widgets": { "application/vnd.jupyter.widget-state+json": { "state": {}, "version_major": 2, "version_minor": 0 } } }, "nbformat": 4, "nbformat_minor": 5 }