id
stringlengths 14
17
| text
stringlengths 42
2.1k
|
|---|---|
4e9727215e95-3100 | Get startedIntroductionInstallationQuickstartModulesModel I/OData connectionChainsMemoryAgentsAgent typesHow-toSubscribing to eventsCancelling requestsCustom LLM AgentCustom LLM Agent (with a ChatModel)Logging and tracingAdding a timeoutToolsToolkitsCallbacksModulesGuidesEcosystemAdditional resourcesCommunity navigatorAPI referenceModulesAgentsHow-toSubscribing to eventsSubscribing to eventsYou can subscribe to a number of events that are emitted by the Agent and the underlying tools, chains and models via callbacks.For more info on the events available see the Callbacks section of the docs.import { initializeAgentExecutorWithOptions } from "langchain/agents";import { OpenAI } from "langchain/llms/openai";import { SerpAPI } from "langchain/tools";import { Calculator } from "langchain/tools/calculator";const model = new OpenAI({ temperature: 0 });const tools = [ new SerpAPI(process.env.SERPAPI_API_KEY, { location: "Austin,Texas,United States", hl: "en", gl: "us", }), new Calculator(),];const executor = await initializeAgentExecutorWithOptions(tools, model, { agentType: "zero-shot-react-description",});const input = `Who is Olivia Wilde's boyfriend? |
4e9727215e95-3101 | What is his current age raised to the 0.23 power?`;const result = await executor.run(input, [ { handleAgentAction(action, runId) { console.log("\nhandleAgentAction", action, runId); }, handleAgentEnd(action, runId) { console.log("\nhandleAgentEnd", action, runId); }, handleToolEnd(output, runId) { console.log("\nhandleToolEnd", output, runId); }, },]);/*handleAgentAction { tool: 'search', toolInput: 'Olivia Wilde boyfriend', log: " I need to find out who Olivia Wilde's boyfriend is and then calculate his age raised to the 0.23 power.\n" + 'Action: search\n' + 'Action Input: "Olivia Wilde boyfriend"'} 9b978461-1f6f-4d5f-80cf-5b229ce181b6handleToolEnd In January 2021, Wilde began dating singer Harry Styles after meeting during the filming of Don't Worry Darling. Their relationship ended in November 2022. |
4e9727215e95-3102 | 062fef47-8ad1-4729-9949-a57be252e002handleAgentAction { tool: 'search', toolInput: 'Harry Styles age', log: " I need to find out Harry Styles' age.\n" + 'Action: search\n' + 'Action Input: "Harry Styles age"'} 9b978461-1f6f-4d5f-80cf-5b229ce181b6handleToolEnd 29 years 9ec91e41-2fbf-4de0-85b6-12b3e6b3784e 61d77e10-c119-435d-a985-1f9d45f0ef08handleAgentAction { tool: 'calculator', toolInput: '29^0.23', log: ' I need to calculate 29 raised to the 0.23 power.\n' + 'Action: calculator\n' + 'Action Input: 29^0.23'} 9b978461-1f6f-4d5f-80cf-5b229ce181b6handleToolEnd 2.169459462491557 07aec96a-ce19-4425-b863-2eae39db8199handleAgentEnd { returnValues: { output: "Harry Styles is Olivia Wilde's boyfriend and his current age raised to the 0.23 power is 2.169459462491557." }, log: ' I now know the final answer.\n' + "Final Answer: Harry Styles is Olivia Wilde's boyfriend and his current age raised to the 0.23 power is 2.169459462491557."} |
4e9727215e95-3103 | age raised to the 0.23 power is 2.169459462491557."} 9b978461-1f6f-4d5f-80cf-5b229ce181b6*/console.log({ result });// { result: "Harry Styles is Olivia Wilde's boyfriend and his current age raised to the 0.23 power is 2.169459462491557." }API Reference:initializeAgentExecutorWithOptions from langchain/agentsOpenAI from langchain/llms/openaiSerpAPI from langchain/toolsCalculator from langchain/tools/calculatorPreviousStructured tool chatNextCancelling requests |
4e9727215e95-3104 | Get startedIntroductionInstallationQuickstartModulesModel I/OData connectionChainsMemoryAgentsAgent typesHow-toSubscribing to eventsCancelling requestsCustom LLM AgentCustom LLM Agent (with a ChatModel)Logging and tracingAdding a timeoutToolsToolkitsCallbacksModulesGuidesEcosystemAdditional resourcesCommunity navigatorAPI reference
ModulesAgentsHow-toSubscribing to eventsSubscribing to eventsYou can subscribe to a number of events that are emitted by the Agent and the underlying tools, chains and models via callbacks.For more info on the events available see the Callbacks section of the docs.import { initializeAgentExecutorWithOptions } from "langchain/agents";import { OpenAI } from "langchain/llms/openai";import { SerpAPI } from "langchain/tools";import { Calculator } from "langchain/tools/calculator";const model = new OpenAI({ temperature: 0 });const tools = [ new SerpAPI(process.env.SERPAPI_API_KEY, { location: "Austin,Texas,United States", hl: "en", gl: "us", }), new Calculator(),];const executor = await initializeAgentExecutorWithOptions(tools, model, { agentType: "zero-shot-react-description",});const input = `Who is Olivia Wilde's boyfriend? |
4e9727215e95-3105 | What is his current age raised to the 0.23 power?`;const result = await executor.run(input, [ { handleAgentAction(action, runId) { console.log("\nhandleAgentAction", action, runId); }, handleAgentEnd(action, runId) { console.log("\nhandleAgentEnd", action, runId); }, handleToolEnd(output, runId) { console.log("\nhandleToolEnd", output, runId); }, },]);/*handleAgentAction { tool: 'search', toolInput: 'Olivia Wilde boyfriend', log: " I need to find out who Olivia Wilde's boyfriend is and then calculate his age raised to the 0.23 power.\n" + 'Action: search\n' + 'Action Input: "Olivia Wilde boyfriend"'} 9b978461-1f6f-4d5f-80cf-5b229ce181b6handleToolEnd In January 2021, Wilde began dating singer Harry Styles after meeting during the filming of Don't Worry Darling. Their relationship ended in November 2022. |
4e9727215e95-3106 | 062fef47-8ad1-4729-9949-a57be252e002handleAgentAction { tool: 'search', toolInput: 'Harry Styles age', log: " I need to find out Harry Styles' age.\n" + 'Action: search\n' + 'Action Input: "Harry Styles age"'} 9b978461-1f6f-4d5f-80cf-5b229ce181b6handleToolEnd 29 years 9ec91e41-2fbf-4de0-85b6-12b3e6b3784e 61d77e10-c119-435d-a985-1f9d45f0ef08handleAgentAction { tool: 'calculator', toolInput: '29^0.23', log: ' I need to calculate 29 raised to the 0.23 power.\n' + 'Action: calculator\n' + 'Action Input: 29^0.23'} 9b978461-1f6f-4d5f-80cf-5b229ce181b6handleToolEnd 2.169459462491557 07aec96a-ce19-4425-b863-2eae39db8199handleAgentEnd { returnValues: { output: "Harry Styles is Olivia Wilde's boyfriend and his current age raised to the 0.23 power is 2.169459462491557." }, log: ' I now know the final answer.\n' + "Final Answer: Harry Styles is Olivia Wilde's boyfriend and his current age raised to the 0.23 power is 2.169459462491557."} |
4e9727215e95-3107 | age raised to the 0.23 power is 2.169459462491557."} 9b978461-1f6f-4d5f-80cf-5b229ce181b6*/console.log({ result });// { result: "Harry Styles is Olivia Wilde's boyfriend and his current age raised to the 0.23 power is 2.169459462491557." }API Reference:initializeAgentExecutorWithOptions from langchain/agentsOpenAI from langchain/llms/openaiSerpAPI from langchain/toolsCalculator from langchain/tools/calculatorPreviousStructured tool chatNextCancelling requests |
4e9727215e95-3108 | Subscribing to eventsYou can subscribe to a number of events that are emitted by the Agent and the underlying tools, chains and models via callbacks.For more info on the events available see the Callbacks section of the docs.import { initializeAgentExecutorWithOptions } from "langchain/agents";import { OpenAI } from "langchain/llms/openai";import { SerpAPI } from "langchain/tools";import { Calculator } from "langchain/tools/calculator";const model = new OpenAI({ temperature: 0 });const tools = [ new SerpAPI(process.env.SERPAPI_API_KEY, { location: "Austin,Texas,United States", hl: "en", gl: "us", }), new Calculator(),];const executor = await initializeAgentExecutorWithOptions(tools, model, { agentType: "zero-shot-react-description",});const input = `Who is Olivia Wilde's boyfriend? |
4e9727215e95-3109 | What is his current age raised to the 0.23 power?`;const result = await executor.run(input, [ { handleAgentAction(action, runId) { console.log("\nhandleAgentAction", action, runId); }, handleAgentEnd(action, runId) { console.log("\nhandleAgentEnd", action, runId); }, handleToolEnd(output, runId) { console.log("\nhandleToolEnd", output, runId); }, },]);/*handleAgentAction { tool: 'search', toolInput: 'Olivia Wilde boyfriend', log: " I need to find out who Olivia Wilde's boyfriend is and then calculate his age raised to the 0.23 power.\n" + 'Action: search\n' + 'Action Input: "Olivia Wilde boyfriend"'} 9b978461-1f6f-4d5f-80cf-5b229ce181b6handleToolEnd In January 2021, Wilde began dating singer Harry Styles after meeting during the filming of Don't Worry Darling. Their relationship ended in November 2022. |
4e9727215e95-3110 | 062fef47-8ad1-4729-9949-a57be252e002handleAgentAction { tool: 'search', toolInput: 'Harry Styles age', log: " I need to find out Harry Styles' age.\n" + 'Action: search\n' + 'Action Input: "Harry Styles age"'} 9b978461-1f6f-4d5f-80cf-5b229ce181b6handleToolEnd 29 years 9ec91e41-2fbf-4de0-85b6-12b3e6b3784e 61d77e10-c119-435d-a985-1f9d45f0ef08handleAgentAction { tool: 'calculator', toolInput: '29^0.23', log: ' I need to calculate 29 raised to the 0.23 power.\n' + 'Action: calculator\n' + 'Action Input: 29^0.23'} 9b978461-1f6f-4d5f-80cf-5b229ce181b6handleToolEnd 2.169459462491557 07aec96a-ce19-4425-b863-2eae39db8199handleAgentEnd { returnValues: { output: "Harry Styles is Olivia Wilde's boyfriend and his current age raised to the 0.23 power is 2.169459462491557." }, log: ' I now know the final answer.\n' + "Final Answer: Harry Styles is Olivia Wilde's boyfriend and his current age raised to the 0.23 power is 2.169459462491557."} |
4e9727215e95-3111 | age raised to the 0.23 power is 2.169459462491557."} 9b978461-1f6f-4d5f-80cf-5b229ce181b6*/console.log({ result });// { result: "Harry Styles is Olivia Wilde's boyfriend and his current age raised to the 0.23 power is 2.169459462491557." }API Reference:initializeAgentExecutorWithOptions from langchain/agentsOpenAI from langchain/llms/openaiSerpAPI from langchain/toolsCalculator from langchain/tools/calculator |
4e9727215e95-3112 | You can subscribe to a number of events that are emitted by the Agent and the underlying tools, chains and models via callbacks. |
4e9727215e95-3113 | import { initializeAgentExecutorWithOptions } from "langchain/agents";import { OpenAI } from "langchain/llms/openai";import { SerpAPI } from "langchain/tools";import { Calculator } from "langchain/tools/calculator";const model = new OpenAI({ temperature: 0 });const tools = [ new SerpAPI(process.env.SERPAPI_API_KEY, { location: "Austin,Texas,United States", hl: "en", gl: "us", }), new Calculator(),];const executor = await initializeAgentExecutorWithOptions(tools, model, { agentType: "zero-shot-react-description",});const input = `Who is Olivia Wilde's boyfriend? What is his current age raised to the 0.23 power?`;const result = await executor.run(input, [ { handleAgentAction(action, runId) { console.log("\nhandleAgentAction", action, runId); }, handleAgentEnd(action, runId) { console.log("\nhandleAgentEnd", action, runId); }, handleToolEnd(output, runId) { console.log("\nhandleToolEnd", output, runId); }, },]);/*handleAgentAction { tool: 'search', toolInput: 'Olivia Wilde boyfriend', log: " I need to find out who Olivia Wilde's boyfriend is and then calculate his age raised to the 0.23 power.\n" + 'Action: search\n' + 'Action Input: "Olivia Wilde boyfriend"'} 9b978461-1f6f-4d5f-80cf-5b229ce181b6handleToolEnd In January 2021, Wilde began dating singer Harry Styles after meeting during the filming of Don't Worry Darling. Their relationship ended in November 2022. |
4e9727215e95-3114 | 062fef47-8ad1-4729-9949-a57be252e002handleAgentAction { tool: 'search', toolInput: 'Harry Styles age', log: " I need to find out Harry Styles' age.\n" + 'Action: search\n' + 'Action Input: "Harry Styles age"'} 9b978461-1f6f-4d5f-80cf-5b229ce181b6handleToolEnd 29 years 9ec91e41-2fbf-4de0-85b6-12b3e6b3784e 61d77e10-c119-435d-a985-1f9d45f0ef08handleAgentAction { tool: 'calculator', toolInput: '29^0.23', log: ' I need to calculate 29 raised to the 0.23 power.\n' + 'Action: calculator\n' + 'Action Input: 29^0.23'} 9b978461-1f6f-4d5f-80cf-5b229ce181b6handleToolEnd 2.169459462491557 07aec96a-ce19-4425-b863-2eae39db8199handleAgentEnd { returnValues: { output: "Harry Styles is Olivia Wilde's boyfriend and his current age raised to the 0.23 power is 2.169459462491557." }, log: ' I now know the final answer.\n' + "Final Answer: Harry Styles is Olivia Wilde's boyfriend and his current age raised to the 0.23 power is 2.169459462491557."} |
4e9727215e95-3115 | age raised to the 0.23 power is 2.169459462491557."} 9b978461-1f6f-4d5f-80cf-5b229ce181b6*/console.log({ result });// { result: "Harry Styles is Olivia Wilde's boyfriend and his current age raised to the 0.23 power is 2.169459462491557." } |
4e9727215e95-3116 | Paragraphs:
Skip to main content🦜️🔗 LangChainDocsUse casesAPILangSmithPython DocsCTRLKGet startedIntroductionInstallationQuickstartModulesModel I/OData connectionChainsMemoryAgentsAgent typesHow-toSubscribing to eventsCancelling requestsCustom LLM AgentCustom LLM Agent (with a ChatModel)Logging and tracingAdding a timeoutToolsToolkitsCallbacksModulesGuidesEcosystemAdditional resourcesCommunity navigatorAPI referenceModulesAgentsHow-toCancelling requestsCancelling requestsYou can cancel a request by passing a signal option when you run the agent. For example:import { initializeAgentExecutorWithOptions } from "langchain/agents";import { OpenAI } from "langchain/llms/openai";import { SerpAPI } from "langchain/tools";import { Calculator } from "langchain/tools/calculator";const model = new OpenAI({ temperature: 0 });const tools = [ new SerpAPI(process.env.SERPAPI_API_KEY, { location: "Austin,Texas,United States", hl: "en", gl: "us", }), new Calculator(),];const executor = await initializeAgentExecutorWithOptions(tools, model, { agentType: "zero-shot-react-description",});const controller = new AbortController();// Call `controller.abort()` somewhere to cancel the request.setTimeout(() => { controller.abort();}, 2000);try { const input = `Who is Olivia Wilde's boyfriend? |
4e9727215e95-3117 | What is his current age raised to the 0.23 power?`; const result = await executor.call({ input, signal: controller.signal });} catch (e) { console.log(e); /* Error: Cancel: canceled at file:///Users/nuno/dev/langchainjs/langchain/dist/util/async_caller.js:60:23 at process.processTicksAndRejections (node:internal/process/task_queues:95:5) at RetryOperation._fn (/Users/nuno/dev/langchainjs/node_modules/p-retry/index.js:50:12) { attemptNumber: 1, retriesLeft: 6 } */}API Reference:initializeAgentExecutorWithOptions from langchain/agentsOpenAI from langchain/llms/openaiSerpAPI from langchain/toolsCalculator from langchain/tools/calculatorNote, this will only cancel the outgoing request if the underlying provider exposes that option. LangChain will cancel the underlying request if possible, otherwise it will cancel the processing of the response.PreviousSubscribing to eventsNextCustom LLM AgentCommunityDiscordTwitterGitHubPythonJS/TSMoreHomepageBlogCopyright © 2023 LangChain, Inc. |
4e9727215e95-3118 | Get startedIntroductionInstallationQuickstartModulesModel I/OData connectionChainsMemoryAgentsAgent typesHow-toSubscribing to eventsCancelling requestsCustom LLM AgentCustom LLM Agent (with a ChatModel)Logging and tracingAdding a timeoutToolsToolkitsCallbacksModulesGuidesEcosystemAdditional resourcesCommunity navigatorAPI referenceModulesAgentsHow-toCancelling requestsCancelling requestsYou can cancel a request by passing a signal option when you run the agent. For example:import { initializeAgentExecutorWithOptions } from "langchain/agents";import { OpenAI } from "langchain/llms/openai";import { SerpAPI } from "langchain/tools";import { Calculator } from "langchain/tools/calculator";const model = new OpenAI({ temperature: 0 });const tools = [ new SerpAPI(process.env.SERPAPI_API_KEY, { location: "Austin,Texas,United States", hl: "en", gl: "us", }), new Calculator(),];const executor = await initializeAgentExecutorWithOptions(tools, model, { agentType: "zero-shot-react-description",});const controller = new AbortController();// Call `controller.abort()` somewhere to cancel the request.setTimeout(() => { controller.abort();}, 2000);try { const input = `Who is Olivia Wilde's boyfriend? |
4e9727215e95-3119 | What is his current age raised to the 0.23 power?`; const result = await executor.call({ input, signal: controller.signal });} catch (e) { console.log(e); /* Error: Cancel: canceled at file:///Users/nuno/dev/langchainjs/langchain/dist/util/async_caller.js:60:23 at process.processTicksAndRejections (node:internal/process/task_queues:95:5) at RetryOperation._fn (/Users/nuno/dev/langchainjs/node_modules/p-retry/index.js:50:12) { attemptNumber: 1, retriesLeft: 6 } */}API Reference:initializeAgentExecutorWithOptions from langchain/agentsOpenAI from langchain/llms/openaiSerpAPI from langchain/toolsCalculator from langchain/tools/calculatorNote, this will only cancel the outgoing request if the underlying provider exposes that option. LangChain will cancel the underlying request if possible, otherwise it will cancel the processing of the response.PreviousSubscribing to eventsNextCustom LLM Agent |
4e9727215e95-3120 | ModulesAgentsHow-toCancelling requestsCancelling requestsYou can cancel a request by passing a signal option when you run the agent. For example:import { initializeAgentExecutorWithOptions } from "langchain/agents";import { OpenAI } from "langchain/llms/openai";import { SerpAPI } from "langchain/tools";import { Calculator } from "langchain/tools/calculator";const model = new OpenAI({ temperature: 0 });const tools = [ new SerpAPI(process.env.SERPAPI_API_KEY, { location: "Austin,Texas,United States", hl: "en", gl: "us", }), new Calculator(),];const executor = await initializeAgentExecutorWithOptions(tools, model, { agentType: "zero-shot-react-description",});const controller = new AbortController();// Call `controller.abort()` somewhere to cancel the request.setTimeout(() => { controller.abort();}, 2000);try { const input = `Who is Olivia Wilde's boyfriend? |
4e9727215e95-3121 | What is his current age raised to the 0.23 power?`; const result = await executor.call({ input, signal: controller.signal });} catch (e) { console.log(e); /* Error: Cancel: canceled at file:///Users/nuno/dev/langchainjs/langchain/dist/util/async_caller.js:60:23 at process.processTicksAndRejections (node:internal/process/task_queues:95:5) at RetryOperation._fn (/Users/nuno/dev/langchainjs/node_modules/p-retry/index.js:50:12) { attemptNumber: 1, retriesLeft: 6 } */}API Reference:initializeAgentExecutorWithOptions from langchain/agentsOpenAI from langchain/llms/openaiSerpAPI from langchain/toolsCalculator from langchain/tools/calculatorNote, this will only cancel the outgoing request if the underlying provider exposes that option. LangChain will cancel the underlying request if possible, otherwise it will cancel the processing of the response.PreviousSubscribing to eventsNextCustom LLM Agent |
4e9727215e95-3122 | Cancelling requestsYou can cancel a request by passing a signal option when you run the agent. For example:import { initializeAgentExecutorWithOptions } from "langchain/agents";import { OpenAI } from "langchain/llms/openai";import { SerpAPI } from "langchain/tools";import { Calculator } from "langchain/tools/calculator";const model = new OpenAI({ temperature: 0 });const tools = [ new SerpAPI(process.env.SERPAPI_API_KEY, { location: "Austin,Texas,United States", hl: "en", gl: "us", }), new Calculator(),];const executor = await initializeAgentExecutorWithOptions(tools, model, { agentType: "zero-shot-react-description",});const controller = new AbortController();// Call `controller.abort()` somewhere to cancel the request.setTimeout(() => { controller.abort();}, 2000);try { const input = `Who is Olivia Wilde's boyfriend? |
4e9727215e95-3123 | What is his current age raised to the 0.23 power?`; const result = await executor.call({ input, signal: controller.signal });} catch (e) { console.log(e); /* Error: Cancel: canceled at file:///Users/nuno/dev/langchainjs/langchain/dist/util/async_caller.js:60:23 at process.processTicksAndRejections (node:internal/process/task_queues:95:5) at RetryOperation._fn (/Users/nuno/dev/langchainjs/node_modules/p-retry/index.js:50:12) { attemptNumber: 1, retriesLeft: 6 } */}API Reference:initializeAgentExecutorWithOptions from langchain/agentsOpenAI from langchain/llms/openaiSerpAPI from langchain/toolsCalculator from langchain/tools/calculatorNote, this will only cancel the outgoing request if the underlying provider exposes that option. LangChain will cancel the underlying request if possible, otherwise it will cancel the processing of the response.
You can cancel a request by passing a signal option when you run the agent. For example: |
4e9727215e95-3124 | You can cancel a request by passing a signal option when you run the agent. For example:
import { initializeAgentExecutorWithOptions } from "langchain/agents";import { OpenAI } from "langchain/llms/openai";import { SerpAPI } from "langchain/tools";import { Calculator } from "langchain/tools/calculator";const model = new OpenAI({ temperature: 0 });const tools = [ new SerpAPI(process.env.SERPAPI_API_KEY, { location: "Austin,Texas,United States", hl: "en", gl: "us", }), new Calculator(),];const executor = await initializeAgentExecutorWithOptions(tools, model, { agentType: "zero-shot-react-description",});const controller = new AbortController();// Call `controller.abort()` somewhere to cancel the request.setTimeout(() => { controller.abort();}, 2000);try { const input = `Who is Olivia Wilde's boyfriend? What is his current age raised to the 0.23 power?`; const result = await executor.call({ input, signal: controller.signal });} catch (e) { console.log(e); /* Error: Cancel: canceled at file:///Users/nuno/dev/langchainjs/langchain/dist/util/async_caller.js:60:23 at process.processTicksAndRejections (node:internal/process/task_queues:95:5) at RetryOperation._fn (/Users/nuno/dev/langchainjs/node_modules/p-retry/index.js:50:12) { attemptNumber: 1, retriesLeft: 6 } */}
Custom LLM Agent
Page Title: Custom LLM Agent | 🦜️🔗 Langchain
Paragraphs: |
4e9727215e95-3125 | Paragraphs:
Skip to main content🦜️🔗 LangChainDocsUse casesAPILangSmithPython DocsCTRLKGet startedIntroductionInstallationQuickstartModulesModel I/OData connectionChainsMemoryAgentsAgent typesHow-toSubscribing to eventsCancelling requestsCustom LLM AgentCustom LLM Agent (with a ChatModel)Logging and tracingAdding a timeoutToolsToolkitsCallbacksModulesGuidesEcosystemAdditional resourcesCommunity navigatorAPI referenceModulesAgentsHow-toCustom LLM AgentCustom LLM AgentThis notebook goes through how to create your own custom LLM agent.An LLM agent consists of three parts:PromptTemplate: This is the prompt template that can be used to instruct the language model on what to doLLM: This is the language model that powers the agentstop sequence: Instructs the LLM to stop generating as soon as this string is foundOutputParser: This determines how to parse the LLMOutput into an AgentAction or AgentFinish objectThe LLMAgent is used in an AgentExecutor. This AgentExecutor can largely be thought of as a loop that:Passes user input and any previous steps to the Agent (in this case, the LLMAgent)If the Agent returns an AgentFinish, then return that directly to the userIf the Agent returns an AgentAction, then use that to call a tool and get an ObservationRepeat, passing the AgentAction and Observation back to the Agent until an AgentFinish is emitted.AgentAction is a response that consists of action and action_input. action refers to which tool to use, and action_input refers to the input to that tool. |
4e9727215e95-3126 | log can also be provided as more context (that can be used for logging, tracing, etc).AgentFinish is a response that contains the final message to be sent back to the user. This should be used to end an agent run.import { LLMSingleActionAgent, AgentActionOutputParser, AgentExecutor,} from "langchain/agents";import { LLMChain } from "langchain/chains";import { OpenAI } from "langchain/llms/openai";import { BasePromptTemplate, BaseStringPromptTemplate, SerializedBasePromptTemplate, renderTemplate,} from "langchain/prompts";import { InputValues, PartialValues, AgentStep, AgentAction, AgentFinish,} from "langchain/schema";import { SerpAPI, Tool } from "langchain/tools";import { Calculator } from "langchain/tools/calculator";const PREFIX = `Answer the following questions as best you can.
You have access to the following tools:`;const formatInstructions = ( toolNames: string) => `Use the following format in your response:Question: the input question you must answerThought: you should always think about what to doAction: the action to take, should be one of [${toolNames}]Action Input: the input to the actionObservation: the result of the action... (this Thought/Action/Action Input/Observation can repeat N times)Thought: I now know the final answerFinal Answer: the final answer to the original input question`;const SUFFIX = `Begin!Question: {input}Thought:{agent_scratchpad}`;class CustomPromptTemplate extends BaseStringPromptTemplate { tools: Tool[]; constructor(args: { tools: Tool[]; inputVariables: string[] }) { super({ inputVariables: args.inputVariables }); this.tools = args.tools; } _getPromptType(): string { |
4e9727215e95-3127 | throw new Error("Not implemented"); } format(input: InputValues): Promise<string> { /** Construct the final template */ const toolStrings = this.tools .map((tool) => `${tool.name}: ${tool.description}`) .join("\n"); const toolNames = this.tools.map((tool) => tool.name).join("\n"); const instructions = formatInstructions(toolNames); const template = [PREFIX, toolStrings, instructions, SUFFIX].join("\n\n"); /** Construct the agent_scratchpad */ const intermediateSteps = input.intermediate_steps as AgentStep[]; const agentScratchpad = intermediateSteps.reduce( (thoughts, { action, observation }) => thoughts + [action.log, `\nObservation: ${observation}`, "Thought:"].join("\n"), "" ); const newInput = { agent_scratchpad: agentScratchpad, ...input }; /** Format the template.
/ return Promise.resolve(renderTemplate(template, "f-string", newInput)); } partial(_values: PartialValues): Promise<BaseStringPromptTemplate> { throw new Error("Not implemented"); } serialize(): SerializedBasePromptTemplate { throw new Error("Not implemented"); }}class CustomOutputParser extends AgentActionOutputParser { lc_namespace = ["langchain", "agents", "custom_llm_agent"]; async parse(text: string): Promise<AgentAction | AgentFinish> { if (text.includes("Final Answer:")) { const parts = text.split("Final Answer:"); const input = parts[parts.length - 1].trim(); const finalAnswers = { output: input }; return { log: text, returnValues: finalAnswers }; } const match = /Action: (. *)\nAction Input: (. |
4e9727215e95-3128 | )/s.exec(text); if (!match) { throw new Error(`Could not parse LLM output: ${text}`); } return { tool: match[1].trim(), toolInput: match[2].trim().replace(/^"+|"+$/g, ""), log: text, }; } getFormatInstructions(): string { throw new Error("Not implemented"); }}export const run = async () => { const model = new OpenAI({ temperature: 0 }); const tools = [ new SerpAPI(process.env.SERPAPI_API_KEY, { location: "Austin,Texas,United States", hl: "en", gl: "us", }), new Calculator(), ]; const llmChain = new LLMChain({ prompt: new CustomPromptTemplate({ tools, inputVariables: ["input", "agent_scratchpad"], }), llm: model, }); const agent = new LLMSingleActionAgent({ llmChain, outputParser: new CustomOutputParser(), stop: ["\nObservation"], }); const executor = new AgentExecutor({ agent, tools, }); console.log("Loaded agent. "); const input = `Who is Olivia Wilde's boyfriend? |
4e9727215e95-3129 | What is his current age raised to the 0.23 power?`; console.log(`Executing with input "${input}"...`); const result = await executor.call({ input }); console.log(`Got output ${result.output}`);};API Reference:LLMSingleActionAgent from langchain/agentsAgentActionOutputParser from langchain/agentsAgentExecutor from langchain/agentsLLMChain from langchain/chainsOpenAI from langchain/llms/openaiBasePromptTemplate from langchain/promptsBaseStringPromptTemplate from langchain/promptsSerializedBasePromptTemplate from langchain/promptsrenderTemplate from langchain/promptsInputValues from langchain/schemaPartialValues from langchain/schemaAgentStep from langchain/schemaAgentAction from langchain/schemaAgentFinish from langchain/schemaSerpAPI from langchain/toolsTool from langchain/toolsCalculator from langchain/tools/calculatorPreviousCancelling requestsNextCustom LLM Agent (with a ChatModel)CommunityDiscordTwitterGitHubPythonJS/TSMoreHomepageBlogCopyright © 2023 LangChain, Inc. |
4e9727215e95-3130 | Get startedIntroductionInstallationQuickstartModulesModel I/OData connectionChainsMemoryAgentsAgent typesHow-toSubscribing to eventsCancelling requestsCustom LLM AgentCustom LLM Agent (with a ChatModel)Logging and tracingAdding a timeoutToolsToolkitsCallbacksModulesGuidesEcosystemAdditional resourcesCommunity navigatorAPI referenceModulesAgentsHow-toCustom LLM AgentCustom LLM AgentThis notebook goes through how to create your own custom LLM agent.An LLM agent consists of three parts:PromptTemplate: This is the prompt template that can be used to instruct the language model on what to doLLM: This is the language model that powers the agentstop sequence: Instructs the LLM to stop generating as soon as this string is foundOutputParser: This determines how to parse the LLMOutput into an AgentAction or AgentFinish objectThe LLMAgent is used in an AgentExecutor. This AgentExecutor can largely be thought of as a loop that:Passes user input and any previous steps to the Agent (in this case, the LLMAgent)If the Agent returns an AgentFinish, then return that directly to the userIf the Agent returns an AgentAction, then use that to call a tool and get an ObservationRepeat, passing the AgentAction and Observation back to the Agent until an AgentFinish is emitted.AgentAction is a response that consists of action and action_input. action refers to which tool to use, and action_input refers to the input to that tool. |
4e9727215e95-3131 | log can also be provided as more context (that can be used for logging, tracing, etc).AgentFinish is a response that contains the final message to be sent back to the user. This should be used to end an agent run.import { LLMSingleActionAgent, AgentActionOutputParser, AgentExecutor,} from "langchain/agents";import { LLMChain } from "langchain/chains";import { OpenAI } from "langchain/llms/openai";import { BasePromptTemplate, BaseStringPromptTemplate, SerializedBasePromptTemplate, renderTemplate,} from "langchain/prompts";import { InputValues, PartialValues, AgentStep, AgentAction, AgentFinish,} from "langchain/schema";import { SerpAPI, Tool } from "langchain/tools";import { Calculator } from "langchain/tools/calculator";const PREFIX = `Answer the following questions as best you can.
You have access to the following tools:`;const formatInstructions = ( toolNames: string) => `Use the following format in your response:Question: the input question you must answerThought: you should always think about what to doAction: the action to take, should be one of [${toolNames}]Action Input: the input to the actionObservation: the result of the action... (this Thought/Action/Action Input/Observation can repeat N times)Thought: I now know the final answerFinal Answer: the final answer to the original input question`;const SUFFIX = `Begin!Question: {input}Thought:{agent_scratchpad}`;class CustomPromptTemplate extends BaseStringPromptTemplate { tools: Tool[]; constructor(args: { tools: Tool[]; inputVariables: string[] }) { super({ inputVariables: args.inputVariables }); this.tools = args.tools; } _getPromptType(): string { |
4e9727215e95-3132 | throw new Error("Not implemented"); } format(input: InputValues): Promise<string> { /** Construct the final template */ const toolStrings = this.tools .map((tool) => `${tool.name}: ${tool.description}`) .join("\n"); const toolNames = this.tools.map((tool) => tool.name).join("\n"); const instructions = formatInstructions(toolNames); const template = [PREFIX, toolStrings, instructions, SUFFIX].join("\n\n"); /** Construct the agent_scratchpad */ const intermediateSteps = input.intermediate_steps as AgentStep[]; const agentScratchpad = intermediateSteps.reduce( (thoughts, { action, observation }) => thoughts + [action.log, `\nObservation: ${observation}`, "Thought:"].join("\n"), "" ); const newInput = { agent_scratchpad: agentScratchpad, ...input }; /** Format the template.
/ return Promise.resolve(renderTemplate(template, "f-string", newInput)); } partial(_values: PartialValues): Promise<BaseStringPromptTemplate> { throw new Error("Not implemented"); } serialize(): SerializedBasePromptTemplate { throw new Error("Not implemented"); }}class CustomOutputParser extends AgentActionOutputParser { lc_namespace = ["langchain", "agents", "custom_llm_agent"]; async parse(text: string): Promise<AgentAction | AgentFinish> { if (text.includes("Final Answer:")) { const parts = text.split("Final Answer:"); const input = parts[parts.length - 1].trim(); const finalAnswers = { output: input }; return { log: text, returnValues: finalAnswers }; } const match = /Action: (. *)\nAction Input: (. |
4e9727215e95-3133 | )/s.exec(text); if (!match) { throw new Error(`Could not parse LLM output: ${text}`); } return { tool: match[1].trim(), toolInput: match[2].trim().replace(/^"+|"+$/g, ""), log: text, }; } getFormatInstructions(): string { throw new Error("Not implemented"); }}export const run = async () => { const model = new OpenAI({ temperature: 0 }); const tools = [ new SerpAPI(process.env.SERPAPI_API_KEY, { location: "Austin,Texas,United States", hl: "en", gl: "us", }), new Calculator(), ]; const llmChain = new LLMChain({ prompt: new CustomPromptTemplate({ tools, inputVariables: ["input", "agent_scratchpad"], }), llm: model, }); const agent = new LLMSingleActionAgent({ llmChain, outputParser: new CustomOutputParser(), stop: ["\nObservation"], }); const executor = new AgentExecutor({ agent, tools, }); console.log("Loaded agent. "); const input = `Who is Olivia Wilde's boyfriend? |
4e9727215e95-3134 | What is his current age raised to the 0.23 power?`; console.log(`Executing with input "${input}"...`); const result = await executor.call({ input }); console.log(`Got output ${result.output}`);};API Reference:LLMSingleActionAgent from langchain/agentsAgentActionOutputParser from langchain/agentsAgentExecutor from langchain/agentsLLMChain from langchain/chainsOpenAI from langchain/llms/openaiBasePromptTemplate from langchain/promptsBaseStringPromptTemplate from langchain/promptsSerializedBasePromptTemplate from langchain/promptsrenderTemplate from langchain/promptsInputValues from langchain/schemaPartialValues from langchain/schemaAgentStep from langchain/schemaAgentAction from langchain/schemaAgentFinish from langchain/schemaSerpAPI from langchain/toolsTool from langchain/toolsCalculator from langchain/tools/calculatorPreviousCancelling requestsNextCustom LLM Agent (with a ChatModel) |
4e9727215e95-3135 | ModulesAgentsHow-toCustom LLM AgentCustom LLM AgentThis notebook goes through how to create your own custom LLM agent.An LLM agent consists of three parts:PromptTemplate: This is the prompt template that can be used to instruct the language model on what to doLLM: This is the language model that powers the agentstop sequence: Instructs the LLM to stop generating as soon as this string is foundOutputParser: This determines how to parse the LLMOutput into an AgentAction or AgentFinish objectThe LLMAgent is used in an AgentExecutor. This AgentExecutor can largely be thought of as a loop that:Passes user input and any previous steps to the Agent (in this case, the LLMAgent)If the Agent returns an AgentFinish, then return that directly to the userIf the Agent returns an AgentAction, then use that to call a tool and get an ObservationRepeat, passing the AgentAction and Observation back to the Agent until an AgentFinish is emitted.AgentAction is a response that consists of action and action_input. action refers to which tool to use, and action_input refers to the input to that tool. log can also be provided as more context (that can be used for logging, tracing, etc).AgentFinish is a response that contains the final message to be sent back to the user. |
4e9727215e95-3136 | This should be used to end an agent run.import { LLMSingleActionAgent, AgentActionOutputParser, AgentExecutor,} from "langchain/agents";import { LLMChain } from "langchain/chains";import { OpenAI } from "langchain/llms/openai";import { BasePromptTemplate, BaseStringPromptTemplate, SerializedBasePromptTemplate, renderTemplate,} from "langchain/prompts";import { InputValues, PartialValues, AgentStep, AgentAction, AgentFinish,} from "langchain/schema";import { SerpAPI, Tool } from "langchain/tools";import { Calculator } from "langchain/tools/calculator";const PREFIX = `Answer the following questions as best you can.
You have access to the following tools:`;const formatInstructions = ( toolNames: string) => `Use the following format in your response:Question: the input question you must answerThought: you should always think about what to doAction: the action to take, should be one of [${toolNames}]Action Input: the input to the actionObservation: the result of the action... (this Thought/Action/Action Input/Observation can repeat N times)Thought: I now know the final answerFinal Answer: the final answer to the original input question`;const SUFFIX = `Begin!Question: {input}Thought:{agent_scratchpad}`;class CustomPromptTemplate extends BaseStringPromptTemplate { tools: Tool[]; constructor(args: { tools: Tool[]; inputVariables: string[] }) { super({ inputVariables: args.inputVariables }); this.tools = args.tools; } _getPromptType(): string { |
4e9727215e95-3137 | throw new Error("Not implemented"); } format(input: InputValues): Promise<string> { /** Construct the final template */ const toolStrings = this.tools .map((tool) => `${tool.name}: ${tool.description}`) .join("\n"); const toolNames = this.tools.map((tool) => tool.name).join("\n"); const instructions = formatInstructions(toolNames); const template = [PREFIX, toolStrings, instructions, SUFFIX].join("\n\n"); /** Construct the agent_scratchpad */ const intermediateSteps = input.intermediate_steps as AgentStep[]; const agentScratchpad = intermediateSteps.reduce( (thoughts, { action, observation }) => thoughts + [action.log, `\nObservation: ${observation}`, "Thought:"].join("\n"), "" ); const newInput = { agent_scratchpad: agentScratchpad, ...input }; /** Format the template.
/ return Promise.resolve(renderTemplate(template, "f-string", newInput)); } partial(_values: PartialValues): Promise<BaseStringPromptTemplate> { throw new Error("Not implemented"); } serialize(): SerializedBasePromptTemplate { throw new Error("Not implemented"); }}class CustomOutputParser extends AgentActionOutputParser { lc_namespace = ["langchain", "agents", "custom_llm_agent"]; async parse(text: string): Promise<AgentAction | AgentFinish> { if (text.includes("Final Answer:")) { const parts = text.split("Final Answer:"); const input = parts[parts.length - 1].trim(); const finalAnswers = { output: input }; return { log: text, returnValues: finalAnswers }; } const match = /Action: (. *)\nAction Input: (. |
4e9727215e95-3138 | )/s.exec(text); if (!match) { throw new Error(`Could not parse LLM output: ${text}`); } return { tool: match[1].trim(), toolInput: match[2].trim().replace(/^"+|"+$/g, ""), log: text, }; } getFormatInstructions(): string { throw new Error("Not implemented"); }}export const run = async () => { const model = new OpenAI({ temperature: 0 }); const tools = [ new SerpAPI(process.env.SERPAPI_API_KEY, { location: "Austin,Texas,United States", hl: "en", gl: "us", }), new Calculator(), ]; const llmChain = new LLMChain({ prompt: new CustomPromptTemplate({ tools, inputVariables: ["input", "agent_scratchpad"], }), llm: model, }); const agent = new LLMSingleActionAgent({ llmChain, outputParser: new CustomOutputParser(), stop: ["\nObservation"], }); const executor = new AgentExecutor({ agent, tools, }); console.log("Loaded agent. "); const input = `Who is Olivia Wilde's boyfriend? |
4e9727215e95-3139 | What is his current age raised to the 0.23 power?`; console.log(`Executing with input "${input}"...`); const result = await executor.call({ input }); console.log(`Got output ${result.output}`);};API Reference:LLMSingleActionAgent from langchain/agentsAgentActionOutputParser from langchain/agentsAgentExecutor from langchain/agentsLLMChain from langchain/chainsOpenAI from langchain/llms/openaiBasePromptTemplate from langchain/promptsBaseStringPromptTemplate from langchain/promptsSerializedBasePromptTemplate from langchain/promptsrenderTemplate from langchain/promptsInputValues from langchain/schemaPartialValues from langchain/schemaAgentStep from langchain/schemaAgentAction from langchain/schemaAgentFinish from langchain/schemaSerpAPI from langchain/toolsTool from langchain/toolsCalculator from langchain/tools/calculatorPreviousCancelling requestsNextCustom LLM Agent (with a ChatModel) |
4e9727215e95-3140 | Custom LLM AgentThis notebook goes through how to create your own custom LLM agent.An LLM agent consists of three parts:PromptTemplate: This is the prompt template that can be used to instruct the language model on what to doLLM: This is the language model that powers the agentstop sequence: Instructs the LLM to stop generating as soon as this string is foundOutputParser: This determines how to parse the LLMOutput into an AgentAction or AgentFinish objectThe LLMAgent is used in an AgentExecutor. This AgentExecutor can largely be thought of as a loop that:Passes user input and any previous steps to the Agent (in this case, the LLMAgent)If the Agent returns an AgentFinish, then return that directly to the userIf the Agent returns an AgentAction, then use that to call a tool and get an ObservationRepeat, passing the AgentAction and Observation back to the Agent until an AgentFinish is emitted.AgentAction is a response that consists of action and action_input. action refers to which tool to use, and action_input refers to the input to that tool. log can also be provided as more context (that can be used for logging, tracing, etc).AgentFinish is a response that contains the final message to be sent back to the user. |
4e9727215e95-3141 | This should be used to end an agent run.import { LLMSingleActionAgent, AgentActionOutputParser, AgentExecutor,} from "langchain/agents";import { LLMChain } from "langchain/chains";import { OpenAI } from "langchain/llms/openai";import { BasePromptTemplate, BaseStringPromptTemplate, SerializedBasePromptTemplate, renderTemplate,} from "langchain/prompts";import { InputValues, PartialValues, AgentStep, AgentAction, AgentFinish,} from "langchain/schema";import { SerpAPI, Tool } from "langchain/tools";import { Calculator } from "langchain/tools/calculator";const PREFIX = `Answer the following questions as best you can.
You have access to the following tools:`;const formatInstructions = ( toolNames: string) => `Use the following format in your response:Question: the input question you must answerThought: you should always think about what to doAction: the action to take, should be one of [${toolNames}]Action Input: the input to the actionObservation: the result of the action... (this Thought/Action/Action Input/Observation can repeat N times)Thought: I now know the final answerFinal Answer: the final answer to the original input question`;const SUFFIX = `Begin!Question: {input}Thought:{agent_scratchpad}`;class CustomPromptTemplate extends BaseStringPromptTemplate { tools: Tool[]; constructor(args: { tools: Tool[]; inputVariables: string[] }) { super({ inputVariables: args.inputVariables }); this.tools = args.tools; } _getPromptType(): string { |
4e9727215e95-3142 | throw new Error("Not implemented"); } format(input: InputValues): Promise<string> { /** Construct the final template */ const toolStrings = this.tools .map((tool) => `${tool.name}: ${tool.description}`) .join("\n"); const toolNames = this.tools.map((tool) => tool.name).join("\n"); const instructions = formatInstructions(toolNames); const template = [PREFIX, toolStrings, instructions, SUFFIX].join("\n\n"); /** Construct the agent_scratchpad */ const intermediateSteps = input.intermediate_steps as AgentStep[]; const agentScratchpad = intermediateSteps.reduce( (thoughts, { action, observation }) => thoughts + [action.log, `\nObservation: ${observation}`, "Thought:"].join("\n"), "" ); const newInput = { agent_scratchpad: agentScratchpad, ...input }; /** Format the template.
/ return Promise.resolve(renderTemplate(template, "f-string", newInput)); } partial(_values: PartialValues): Promise<BaseStringPromptTemplate> { throw new Error("Not implemented"); } serialize(): SerializedBasePromptTemplate { throw new Error("Not implemented"); }}class CustomOutputParser extends AgentActionOutputParser { lc_namespace = ["langchain", "agents", "custom_llm_agent"]; async parse(text: string): Promise<AgentAction | AgentFinish> { if (text.includes("Final Answer:")) { const parts = text.split("Final Answer:"); const input = parts[parts.length - 1].trim(); const finalAnswers = { output: input }; return { log: text, returnValues: finalAnswers }; } const match = /Action: (. *)\nAction Input: (. |
4e9727215e95-3143 | )/s.exec(text); if (!match) { throw new Error(`Could not parse LLM output: ${text}`); } return { tool: match[1].trim(), toolInput: match[2].trim().replace(/^"+|"+$/g, ""), log: text, }; } getFormatInstructions(): string { throw new Error("Not implemented"); }}export const run = async () => { const model = new OpenAI({ temperature: 0 }); const tools = [ new SerpAPI(process.env.SERPAPI_API_KEY, { location: "Austin,Texas,United States", hl: "en", gl: "us", }), new Calculator(), ]; const llmChain = new LLMChain({ prompt: new CustomPromptTemplate({ tools, inputVariables: ["input", "agent_scratchpad"], }), llm: model, }); const agent = new LLMSingleActionAgent({ llmChain, outputParser: new CustomOutputParser(), stop: ["\nObservation"], }); const executor = new AgentExecutor({ agent, tools, }); console.log("Loaded agent. "); const input = `Who is Olivia Wilde's boyfriend? |
4e9727215e95-3144 | What is his current age raised to the 0.23 power?`; console.log(`Executing with input "${input}"...`); const result = await executor.call({ input }); console.log(`Got output ${result.output}`);};API Reference:LLMSingleActionAgent from langchain/agentsAgentActionOutputParser from langchain/agentsAgentExecutor from langchain/agentsLLMChain from langchain/chainsOpenAI from langchain/llms/openaiBasePromptTemplate from langchain/promptsBaseStringPromptTemplate from langchain/promptsSerializedBasePromptTemplate from langchain/promptsrenderTemplate from langchain/promptsInputValues from langchain/schemaPartialValues from langchain/schemaAgentStep from langchain/schemaAgentAction from langchain/schemaAgentFinish from langchain/schemaSerpAPI from langchain/toolsTool from langchain/toolsCalculator from langchain/tools/calculator
This notebook goes through how to create your own custom LLM agent.
An LLM agent consists of three parts:
The LLMAgent is used in an AgentExecutor. This AgentExecutor can largely be thought of as a loop that:
AgentAction is a response that consists of action and action_input. action refers to which tool to use, and action_input refers to the input to that tool. log can also be provided as more context (that can be used for logging, tracing, etc).
AgentFinish is a response that contains the final message to be sent back to the user. This should be used to end an agent run. |
4e9727215e95-3145 | import { LLMSingleActionAgent, AgentActionOutputParser, AgentExecutor,} from "langchain/agents";import { LLMChain } from "langchain/chains";import { OpenAI } from "langchain/llms/openai";import { BasePromptTemplate, BaseStringPromptTemplate, SerializedBasePromptTemplate, renderTemplate,} from "langchain/prompts";import { InputValues, PartialValues, AgentStep, AgentAction, AgentFinish,} from "langchain/schema";import { SerpAPI, Tool } from "langchain/tools";import { Calculator } from "langchain/tools/calculator";const PREFIX = `Answer the following questions as best you can.
You have access to the following tools:`;const formatInstructions = ( toolNames: string) => `Use the following format in your response:Question: the input question you must answerThought: you should always think about what to doAction: the action to take, should be one of [${toolNames}]Action Input: the input to the actionObservation: the result of the action... (this Thought/Action/Action Input/Observation can repeat N times)Thought: I now know the final answerFinal Answer: the final answer to the original input question`;const SUFFIX = `Begin!Question: {input}Thought:{agent_scratchpad}`;class CustomPromptTemplate extends BaseStringPromptTemplate { tools: Tool[]; constructor(args: { tools: Tool[]; inputVariables: string[] }) { super({ inputVariables: args.inputVariables }); this.tools = args.tools; } _getPromptType(): string { |
4e9727215e95-3146 | throw new Error("Not implemented"); } format(input: InputValues): Promise<string> { /** Construct the final template */ const toolStrings = this.tools .map((tool) => `${tool.name}: ${tool.description}`) .join("\n"); const toolNames = this.tools.map((tool) => tool.name).join("\n"); const instructions = formatInstructions(toolNames); const template = [PREFIX, toolStrings, instructions, SUFFIX].join("\n\n"); /** Construct the agent_scratchpad */ const intermediateSteps = input.intermediate_steps as AgentStep[]; const agentScratchpad = intermediateSteps.reduce( (thoughts, { action, observation }) => thoughts + [action.log, `\nObservation: ${observation}`, "Thought:"].join("\n"), "" ); const newInput = { agent_scratchpad: agentScratchpad, ...input }; /** Format the template.
/ return Promise.resolve(renderTemplate(template, "f-string", newInput)); } partial(_values: PartialValues): Promise<BaseStringPromptTemplate> { throw new Error("Not implemented"); } serialize(): SerializedBasePromptTemplate { throw new Error("Not implemented"); }}class CustomOutputParser extends AgentActionOutputParser { lc_namespace = ["langchain", "agents", "custom_llm_agent"]; async parse(text: string): Promise<AgentAction | AgentFinish> { if (text.includes("Final Answer:")) { const parts = text.split("Final Answer:"); const input = parts[parts.length - 1].trim(); const finalAnswers = { output: input }; return { log: text, returnValues: finalAnswers }; } const match = /Action: (. *)\nAction Input: (. |
4e9727215e95-3147 | )/s.exec(text); if (!match) { throw new Error(`Could not parse LLM output: ${text}`); } return { tool: match[1].trim(), toolInput: match[2].trim().replace(/^"+|"+$/g, ""), log: text, }; } getFormatInstructions(): string { throw new Error("Not implemented"); }}export const run = async () => { const model = new OpenAI({ temperature: 0 }); const tools = [ new SerpAPI(process.env.SERPAPI_API_KEY, { location: "Austin,Texas,United States", hl: "en", gl: "us", }), new Calculator(), ]; const llmChain = new LLMChain({ prompt: new CustomPromptTemplate({ tools, inputVariables: ["input", "agent_scratchpad"], }), llm: model, }); const agent = new LLMSingleActionAgent({ llmChain, outputParser: new CustomOutputParser(), stop: ["\nObservation"], }); const executor = new AgentExecutor({ agent, tools, }); console.log("Loaded agent. "); const input = `Who is Olivia Wilde's boyfriend? What is his current age raised to the 0.23 power?`; console.log(`Executing with input "${input}"...`); const result = await executor.call({ input }); console.log(`Got output ${result.output}`);}; |
4e9727215e95-3148 | API Reference:LLMSingleActionAgent from langchain/agentsAgentActionOutputParser from langchain/agentsAgentExecutor from langchain/agentsLLMChain from langchain/chainsOpenAI from langchain/llms/openaiBasePromptTemplate from langchain/promptsBaseStringPromptTemplate from langchain/promptsSerializedBasePromptTemplate from langchain/promptsrenderTemplate from langchain/promptsInputValues from langchain/schemaPartialValues from langchain/schemaAgentStep from langchain/schemaAgentAction from langchain/schemaAgentFinish from langchain/schemaSerpAPI from langchain/toolsTool from langchain/toolsCalculator from langchain/tools/calculator
Custom LLM Agent (with a ChatModel)
Page Title: Custom LLM Agent (with a ChatModel) | 🦜️🔗 Langchain
Paragraphs: |
4e9727215e95-3149 | Paragraphs:
Skip to main content🦜️🔗 LangChainDocsUse casesAPILangSmithPython DocsCTRLKGet startedIntroductionInstallationQuickstartModulesModel I/OData connectionChainsMemoryAgentsAgent typesHow-toSubscribing to eventsCancelling requestsCustom LLM AgentCustom LLM Agent (with a ChatModel)Logging and tracingAdding a timeoutToolsToolkitsCallbacksModulesGuidesEcosystemAdditional resourcesCommunity navigatorAPI referenceModulesAgentsHow-toCustom LLM Agent (with a ChatModel)Custom LLM Agent (with a ChatModel)This notebook goes through how to create your own custom agent based on a chat model.An LLM chat agent consists of three parts:PromptTemplate: This is the prompt template that can be used to instruct the language model on what to doChatModel: This is the language model that powers the agentstop sequence: Instructs the LLM to stop generating as soon as this string is foundOutputParser: This determines how to parse the LLMOutput into an AgentAction or AgentFinish objectThe LLMAgent is used in an AgentExecutor. This AgentExecutor can largely be thought of as a loop that:Passes user input and any previous steps to the Agent (in this case, the LLMAgent)If the Agent returns an AgentFinish, then return that directly to the userIf the Agent returns an AgentAction, then use that to call a tool and get an ObservationRepeat, passing the AgentAction and Observation back to the Agent until an AgentFinish is emitted.AgentAction is a response that consists of action and action_input. |
4e9727215e95-3150 | action refers to which tool to use, and action_input refers to the input to that tool. log can also be provided as more context (that can be used for logging, tracing, etc).AgentFinish is a response that contains the final message to be sent back to the user. This should be used to end an agent run.import { AgentActionOutputParser, AgentExecutor, LLMSingleActionAgent,} from "langchain/agents";import { LLMChain } from "langchain/chains";import { ChatOpenAI } from "langchain/chat_models/openai";import { BaseChatPromptTemplate, BasePromptTemplate, SerializedBasePromptTemplate, renderTemplate,} from "langchain/prompts";import { AgentAction, AgentFinish, AgentStep, BaseMessage, HumanMessage, InputValues, PartialValues,} from "langchain/schema";import { SerpAPI, Tool } from "langchain/tools";import { Calculator } from "langchain/tools/calculator";const PREFIX = `Answer the following questions as best you can. |
4e9727215e95-3151 | You have access to the following tools:`;const formatInstructions = ( toolNames: string) => `Use the following format in your response:Question: the input question you must answerThought: you should always think about what to doAction: the action to take, should be one of [${toolNames}]Action Input: the input to the actionObservation: the result of the action... (this Thought/Action/Action Input/Observation can repeat N times)Thought: I now know the final answerFinal Answer: the final answer to the original input question`;const SUFFIX = `Begin!Question: {input}Thought:{agent_scratchpad}`;class CustomPromptTemplate extends BaseChatPromptTemplate { tools: Tool[]; constructor(args: { tools: Tool[]; inputVariables: string[] }) { super({ inputVariables: args.inputVariables }); this.tools = args.tools; } _getPromptType(): string { throw new |
4e9727215e95-3152 | Error("Not implemented"); } async formatMessages(values: InputValues): Promise<BaseMessage[]> { /** Construct the final template */ const toolStrings = this.tools .map((tool) => `${tool.name}: ${tool.description}`) .join("\n"); const toolNames = this.tools.map((tool) => tool.name).join("\n"); const instructions = formatInstructions(toolNames); const template = [PREFIX, toolStrings, instructions, SUFFIX].join("\n\n"); /** Construct the agent_scratchpad */ const intermediateSteps = values.intermediate_steps as AgentStep[]; const agentScratchpad = intermediateSteps.reduce( (thoughts, { action, observation }) => thoughts + [action.log, `\nObservation: ${observation}`, "Thought:"].join("\n"), "" ); const newInput = { agent_scratchpad: agentScratchpad, ...values }; /** Format the template.
/ const formatted = renderTemplate(template, "f-string", newInput); return [new HumanMessage(formatted)]; } partial(_values: PartialValues): Promise<BaseChatPromptTemplate> { throw new Error("Not implemented"); } serialize(): SerializedBasePromptTemplate { throw new Error("Not implemented"); }}class CustomOutputParser extends AgentActionOutputParser { lc_namespace = ["langchain", "agents", "custom_llm_agent_chat"]; async parse(text: string): Promise<AgentAction | AgentFinish> { if (text.includes("Final Answer:")) { const parts = text.split("Final Answer:"); const input = parts[parts.length - 1].trim(); const finalAnswers = { output: input }; return { log: text, returnValues: finalAnswers }; } const match = /Action: (. *)\nAction Input: (. |
4e9727215e95-3153 | )/s.exec(text); if (!match) { throw new Error(`Could not parse LLM output: ${text}`); } return { tool: match[1].trim(), toolInput: match[2].trim().replace(/^"+|"+$/g, ""), log: text, }; } getFormatInstructions(): string { throw new Error("Not implemented"); }}export const run = async () => { const model = new ChatOpenAI({ temperature: 0 }); const tools = [ new SerpAPI(process.env.SERPAPI_API_KEY, { location: "Austin,Texas,United States", hl: "en", gl: "us", }), new Calculator(), ]; const llmChain = new LLMChain({ prompt: new CustomPromptTemplate({ tools, inputVariables: ["input", "agent_scratchpad"], }), llm: model, }); const agent = new LLMSingleActionAgent({ llmChain, outputParser: new CustomOutputParser(), stop: ["\nObservation"], }); const executor = new AgentExecutor({ agent, tools, }); console.log("Loaded agent. "); const input = `Who is Olivia Wilde's boyfriend? |
4e9727215e95-3154 | What is his current age raised to the 0.23 power?`; console.log(`Executing with input "${input}"...`); const result = await executor.call({ input }); console.log(`Got output ${result.output}`);};run();API Reference:AgentActionOutputParser from langchain/agentsAgentExecutor from langchain/agentsLLMSingleActionAgent from langchain/agentsLLMChain from langchain/chainsChatOpenAI from langchain/chat_models/openaiBaseChatPromptTemplate from langchain/promptsBasePromptTemplate from langchain/promptsSerializedBasePromptTemplate from langchain/promptsrenderTemplate from langchain/promptsAgentAction from langchain/schemaAgentFinish from langchain/schemaAgentStep from langchain/schemaBaseMessage from langchain/schemaHumanMessage from langchain/schemaInputValues from langchain/schemaPartialValues from langchain/schemaSerpAPI from langchain/toolsTool from langchain/toolsCalculator from langchain/tools/calculatorPreviousCustom LLM AgentNextLogging and tracingCommunityDiscordTwitterGitHubPythonJS/TSMoreHomepageBlogCopyright © 2023 LangChain, Inc. |
4e9727215e95-3155 | Get startedIntroductionInstallationQuickstartModulesModel I/OData connectionChainsMemoryAgentsAgent typesHow-toSubscribing to eventsCancelling requestsCustom LLM AgentCustom LLM Agent (with a ChatModel)Logging and tracingAdding a timeoutToolsToolkitsCallbacksModulesGuidesEcosystemAdditional resourcesCommunity navigatorAPI referenceModulesAgentsHow-toCustom LLM Agent (with a ChatModel)Custom LLM Agent (with a ChatModel)This notebook goes through how to create your own custom agent based on a chat model.An LLM chat agent consists of three parts:PromptTemplate: This is the prompt template that can be used to instruct the language model on what to doChatModel: This is the language model that powers the agentstop sequence: Instructs the LLM to stop generating as soon as this string is foundOutputParser: This determines how to parse the LLMOutput into an AgentAction or AgentFinish objectThe LLMAgent is used in an AgentExecutor. This AgentExecutor can largely be thought of as a loop that:Passes user input and any previous steps to the Agent (in this case, the LLMAgent)If the Agent returns an AgentFinish, then return that directly to the userIf the Agent returns an AgentAction, then use that to call a tool and get an ObservationRepeat, passing the AgentAction and Observation back to the Agent until an AgentFinish is emitted.AgentAction is a response that consists of action and action_input. action refers to which tool to use, and action_input refers to the input to that tool. |
4e9727215e95-3156 | log can also be provided as more context (that can be used for logging, tracing, etc).AgentFinish is a response that contains the final message to be sent back to the user. This should be used to end an agent run.import { AgentActionOutputParser, AgentExecutor, LLMSingleActionAgent,} from "langchain/agents";import { LLMChain } from "langchain/chains";import { ChatOpenAI } from "langchain/chat_models/openai";import { BaseChatPromptTemplate, BasePromptTemplate, SerializedBasePromptTemplate, renderTemplate,} from "langchain/prompts";import { AgentAction, AgentFinish, AgentStep, BaseMessage, HumanMessage, InputValues, PartialValues,} from "langchain/schema";import { SerpAPI, Tool } from "langchain/tools";import { Calculator } from "langchain/tools/calculator";const PREFIX = `Answer the following questions as best you can. |
4e9727215e95-3157 | You have access to the following tools:`;const formatInstructions = ( toolNames: string) => `Use the following format in your response:Question: the input question you must answerThought: you should always think about what to doAction: the action to take, should be one of [${toolNames}]Action Input: the input to the actionObservation: the result of the action... (this Thought/Action/Action Input/Observation can repeat N times)Thought: I now know the final answerFinal Answer: the final answer to the original input question`;const SUFFIX = `Begin!Question: {input}Thought:{agent_scratchpad}`;class CustomPromptTemplate extends BaseChatPromptTemplate { tools: Tool[]; constructor(args: { tools: Tool[]; inputVariables: string[] }) { super({ inputVariables: args.inputVariables }); this.tools = args.tools; } _getPromptType(): string { throw new |
4e9727215e95-3158 | Error("Not implemented"); } async formatMessages(values: InputValues): Promise<BaseMessage[]> { /** Construct the final template */ const toolStrings = this.tools .map((tool) => `${tool.name}: ${tool.description}`) .join("\n"); const toolNames = this.tools.map((tool) => tool.name).join("\n"); const instructions = formatInstructions(toolNames); const template = [PREFIX, toolStrings, instructions, SUFFIX].join("\n\n"); /** Construct the agent_scratchpad */ const intermediateSteps = values.intermediate_steps as AgentStep[]; const agentScratchpad = intermediateSteps.reduce( (thoughts, { action, observation }) => thoughts + [action.log, `\nObservation: ${observation}`, "Thought:"].join("\n"), "" ); const newInput = { agent_scratchpad: agentScratchpad, ...values }; /** Format the template.
/ const formatted = renderTemplate(template, "f-string", newInput); return [new HumanMessage(formatted)]; } partial(_values: PartialValues): Promise<BaseChatPromptTemplate> { throw new Error("Not implemented"); } serialize(): SerializedBasePromptTemplate { throw new Error("Not implemented"); }}class CustomOutputParser extends AgentActionOutputParser { lc_namespace = ["langchain", "agents", "custom_llm_agent_chat"]; async parse(text: string): Promise<AgentAction | AgentFinish> { if (text.includes("Final Answer:")) { const parts = text.split("Final Answer:"); const input = parts[parts.length - 1].trim(); const finalAnswers = { output: input }; return { log: text, returnValues: finalAnswers }; } const match = /Action: (. *)\nAction Input: (. |
4e9727215e95-3159 | )/s.exec(text); if (!match) { throw new Error(`Could not parse LLM output: ${text}`); } return { tool: match[1].trim(), toolInput: match[2].trim().replace(/^"+|"+$/g, ""), log: text, }; } getFormatInstructions(): string { throw new Error("Not implemented"); }}export const run = async () => { const model = new ChatOpenAI({ temperature: 0 }); const tools = [ new SerpAPI(process.env.SERPAPI_API_KEY, { location: "Austin,Texas,United States", hl: "en", gl: "us", }), new Calculator(), ]; const llmChain = new LLMChain({ prompt: new CustomPromptTemplate({ tools, inputVariables: ["input", "agent_scratchpad"], }), llm: model, }); const agent = new LLMSingleActionAgent({ llmChain, outputParser: new CustomOutputParser(), stop: ["\nObservation"], }); const executor = new AgentExecutor({ agent, tools, }); console.log("Loaded agent. "); const input = `Who is Olivia Wilde's boyfriend? |
4e9727215e95-3160 | What is his current age raised to the 0.23 power?`; console.log(`Executing with input "${input}"...`); const result = await executor.call({ input }); console.log(`Got output ${result.output}`);};run();API Reference:AgentActionOutputParser from langchain/agentsAgentExecutor from langchain/agentsLLMSingleActionAgent from langchain/agentsLLMChain from langchain/chainsChatOpenAI from langchain/chat_models/openaiBaseChatPromptTemplate from langchain/promptsBasePromptTemplate from langchain/promptsSerializedBasePromptTemplate from langchain/promptsrenderTemplate from langchain/promptsAgentAction from langchain/schemaAgentFinish from langchain/schemaAgentStep from langchain/schemaBaseMessage from langchain/schemaHumanMessage from langchain/schemaInputValues from langchain/schemaPartialValues from langchain/schemaSerpAPI from langchain/toolsTool from langchain/toolsCalculator from langchain/tools/calculatorPreviousCustom LLM AgentNextLogging and tracing |
4e9727215e95-3161 | ModulesAgentsHow-toCustom LLM Agent (with a ChatModel)Custom LLM Agent (with a ChatModel)This notebook goes through how to create your own custom agent based on a chat model.An LLM chat agent consists of three parts:PromptTemplate: This is the prompt template that can be used to instruct the language model on what to doChatModel: This is the language model that powers the agentstop sequence: Instructs the LLM to stop generating as soon as this string is foundOutputParser: This determines how to parse the LLMOutput into an AgentAction or AgentFinish objectThe LLMAgent is used in an AgentExecutor. This AgentExecutor can largely be thought of as a loop that:Passes user input and any previous steps to the Agent (in this case, the LLMAgent)If the Agent returns an AgentFinish, then return that directly to the userIf the Agent returns an AgentAction, then use that to call a tool and get an ObservationRepeat, passing the AgentAction and Observation back to the Agent until an AgentFinish is emitted.AgentAction is a response that consists of action and action_input. action refers to which tool to use, and action_input refers to the input to that tool. log can also be provided as more context (that can be used for logging, tracing, etc).AgentFinish is a response that contains the final message to be sent back to the user. |
4e9727215e95-3162 | This should be used to end an agent run.import { AgentActionOutputParser, AgentExecutor, LLMSingleActionAgent,} from "langchain/agents";import { LLMChain } from "langchain/chains";import { ChatOpenAI } from "langchain/chat_models/openai";import { BaseChatPromptTemplate, BasePromptTemplate, SerializedBasePromptTemplate, renderTemplate,} from "langchain/prompts";import { AgentAction, AgentFinish, AgentStep, BaseMessage, HumanMessage, InputValues, PartialValues,} from "langchain/schema";import { SerpAPI, Tool } from "langchain/tools";import { Calculator } from "langchain/tools/calculator";const PREFIX = `Answer the following questions as best you can.
You have access to the following tools:`;const formatInstructions = ( toolNames: string) => `Use the following format in your response:Question: the input question you must answerThought: you should always think about what to doAction: the action to take, should be one of [${toolNames}]Action Input: the input to the actionObservation: the result of the action... (this Thought/Action/Action Input/Observation can repeat N times)Thought: I now know the final answerFinal Answer: the final answer to the original input question`;const SUFFIX = `Begin!Question: {input}Thought:{agent_scratchpad}`;class CustomPromptTemplate extends BaseChatPromptTemplate { tools: Tool[]; constructor(args: { tools: Tool[]; inputVariables: string[] }) { super({ inputVariables: args.inputVariables }); this.tools = args.tools; } _getPromptType(): string { throw new |
4e9727215e95-3163 | Error("Not implemented"); } async formatMessages(values: InputValues): Promise<BaseMessage[]> { /** Construct the final template */ const toolStrings = this.tools .map((tool) => `${tool.name}: ${tool.description}`) .join("\n"); const toolNames = this.tools.map((tool) => tool.name).join("\n"); const instructions = formatInstructions(toolNames); const template = [PREFIX, toolStrings, instructions, SUFFIX].join("\n\n"); /** Construct the agent_scratchpad */ const intermediateSteps = values.intermediate_steps as AgentStep[]; const agentScratchpad = intermediateSteps.reduce( (thoughts, { action, observation }) => thoughts + [action.log, `\nObservation: ${observation}`, "Thought:"].join("\n"), "" ); const newInput = { agent_scratchpad: agentScratchpad, ...values }; /** Format the template.
/ const formatted = renderTemplate(template, "f-string", newInput); return [new HumanMessage(formatted)]; } partial(_values: PartialValues): Promise<BaseChatPromptTemplate> { throw new Error("Not implemented"); } serialize(): SerializedBasePromptTemplate { throw new Error("Not implemented"); }}class CustomOutputParser extends AgentActionOutputParser { lc_namespace = ["langchain", "agents", "custom_llm_agent_chat"]; async parse(text: string): Promise<AgentAction | AgentFinish> { if (text.includes("Final Answer:")) { const parts = text.split("Final Answer:"); const input = parts[parts.length - 1].trim(); const finalAnswers = { output: input }; return { log: text, returnValues: finalAnswers }; } const match = /Action: (. *)\nAction Input: (. |
4e9727215e95-3164 | )/s.exec(text); if (!match) { throw new Error(`Could not parse LLM output: ${text}`); } return { tool: match[1].trim(), toolInput: match[2].trim().replace(/^"+|"+$/g, ""), log: text, }; } getFormatInstructions(): string { throw new Error("Not implemented"); }}export const run = async () => { const model = new ChatOpenAI({ temperature: 0 }); const tools = [ new SerpAPI(process.env.SERPAPI_API_KEY, { location: "Austin,Texas,United States", hl: "en", gl: "us", }), new Calculator(), ]; const llmChain = new LLMChain({ prompt: new CustomPromptTemplate({ tools, inputVariables: ["input", "agent_scratchpad"], }), llm: model, }); const agent = new LLMSingleActionAgent({ llmChain, outputParser: new CustomOutputParser(), stop: ["\nObservation"], }); const executor = new AgentExecutor({ agent, tools, }); console.log("Loaded agent. "); const input = `Who is Olivia Wilde's boyfriend? |
4e9727215e95-3165 | What is his current age raised to the 0.23 power?`; console.log(`Executing with input "${input}"...`); const result = await executor.call({ input }); console.log(`Got output ${result.output}`);};run();API Reference:AgentActionOutputParser from langchain/agentsAgentExecutor from langchain/agentsLLMSingleActionAgent from langchain/agentsLLMChain from langchain/chainsChatOpenAI from langchain/chat_models/openaiBaseChatPromptTemplate from langchain/promptsBasePromptTemplate from langchain/promptsSerializedBasePromptTemplate from langchain/promptsrenderTemplate from langchain/promptsAgentAction from langchain/schemaAgentFinish from langchain/schemaAgentStep from langchain/schemaBaseMessage from langchain/schemaHumanMessage from langchain/schemaInputValues from langchain/schemaPartialValues from langchain/schemaSerpAPI from langchain/toolsTool from langchain/toolsCalculator from langchain/tools/calculatorPreviousCustom LLM AgentNextLogging and tracing |
4e9727215e95-3166 | Custom LLM Agent (with a ChatModel)This notebook goes through how to create your own custom agent based on a chat model.An LLM chat agent consists of three parts:PromptTemplate: This is the prompt template that can be used to instruct the language model on what to doChatModel: This is the language model that powers the agentstop sequence: Instructs the LLM to stop generating as soon as this string is foundOutputParser: This determines how to parse the LLMOutput into an AgentAction or AgentFinish objectThe LLMAgent is used in an AgentExecutor. This AgentExecutor can largely be thought of as a loop that:Passes user input and any previous steps to the Agent (in this case, the LLMAgent)If the Agent returns an AgentFinish, then return that directly to the userIf the Agent returns an AgentAction, then use that to call a tool and get an ObservationRepeat, passing the AgentAction and Observation back to the Agent until an AgentFinish is emitted.AgentAction is a response that consists of action and action_input. action refers to which tool to use, and action_input refers to the input to that tool. log can also be provided as more context (that can be used for logging, tracing, etc).AgentFinish is a response that contains the final message to be sent back to the user. |
4e9727215e95-3167 | This should be used to end an agent run.import { AgentActionOutputParser, AgentExecutor, LLMSingleActionAgent,} from "langchain/agents";import { LLMChain } from "langchain/chains";import { ChatOpenAI } from "langchain/chat_models/openai";import { BaseChatPromptTemplate, BasePromptTemplate, SerializedBasePromptTemplate, renderTemplate,} from "langchain/prompts";import { AgentAction, AgentFinish, AgentStep, BaseMessage, HumanMessage, InputValues, PartialValues,} from "langchain/schema";import { SerpAPI, Tool } from "langchain/tools";import { Calculator } from "langchain/tools/calculator";const PREFIX = `Answer the following questions as best you can.
You have access to the following tools:`;const formatInstructions = ( toolNames: string) => `Use the following format in your response:Question: the input question you must answerThought: you should always think about what to doAction: the action to take, should be one of [${toolNames}]Action Input: the input to the actionObservation: the result of the action... (this Thought/Action/Action Input/Observation can repeat N times)Thought: I now know the final answerFinal Answer: the final answer to the original input question`;const SUFFIX = `Begin!Question: {input}Thought:{agent_scratchpad}`;class CustomPromptTemplate extends BaseChatPromptTemplate { tools: Tool[]; constructor(args: { tools: Tool[]; inputVariables: string[] }) { super({ inputVariables: args.inputVariables }); this.tools = args.tools; } _getPromptType(): string { throw new |
4e9727215e95-3168 | Error("Not implemented"); } async formatMessages(values: InputValues): Promise<BaseMessage[]> { /** Construct the final template */ const toolStrings = this.tools .map((tool) => `${tool.name}: ${tool.description}`) .join("\n"); const toolNames = this.tools.map((tool) => tool.name).join("\n"); const instructions = formatInstructions(toolNames); const template = [PREFIX, toolStrings, instructions, SUFFIX].join("\n\n"); /** Construct the agent_scratchpad */ const intermediateSteps = values.intermediate_steps as AgentStep[]; const agentScratchpad = intermediateSteps.reduce( (thoughts, { action, observation }) => thoughts + [action.log, `\nObservation: ${observation}`, "Thought:"].join("\n"), "" ); const newInput = { agent_scratchpad: agentScratchpad, ...values }; /** Format the template.
/ const formatted = renderTemplate(template, "f-string", newInput); return [new HumanMessage(formatted)]; } partial(_values: PartialValues): Promise<BaseChatPromptTemplate> { throw new Error("Not implemented"); } serialize(): SerializedBasePromptTemplate { throw new Error("Not implemented"); }}class CustomOutputParser extends AgentActionOutputParser { lc_namespace = ["langchain", "agents", "custom_llm_agent_chat"]; async parse(text: string): Promise<AgentAction | AgentFinish> { if (text.includes("Final Answer:")) { const parts = text.split("Final Answer:"); const input = parts[parts.length - 1].trim(); const finalAnswers = { output: input }; return { log: text, returnValues: finalAnswers }; } const match = /Action: (. *)\nAction Input: (. |
4e9727215e95-3169 | )/s.exec(text); if (!match) { throw new Error(`Could not parse LLM output: ${text}`); } return { tool: match[1].trim(), toolInput: match[2].trim().replace(/^"+|"+$/g, ""), log: text, }; } getFormatInstructions(): string { throw new Error("Not implemented"); }}export const run = async () => { const model = new ChatOpenAI({ temperature: 0 }); const tools = [ new SerpAPI(process.env.SERPAPI_API_KEY, { location: "Austin,Texas,United States", hl: "en", gl: "us", }), new Calculator(), ]; const llmChain = new LLMChain({ prompt: new CustomPromptTemplate({ tools, inputVariables: ["input", "agent_scratchpad"], }), llm: model, }); const agent = new LLMSingleActionAgent({ llmChain, outputParser: new CustomOutputParser(), stop: ["\nObservation"], }); const executor = new AgentExecutor({ agent, tools, }); console.log("Loaded agent. "); const input = `Who is Olivia Wilde's boyfriend? |
4e9727215e95-3170 | What is his current age raised to the 0.23 power?`; console.log(`Executing with input "${input}"...`); const result = await executor.call({ input }); console.log(`Got output ${result.output}`);};run();API Reference:AgentActionOutputParser from langchain/agentsAgentExecutor from langchain/agentsLLMSingleActionAgent from langchain/agentsLLMChain from langchain/chainsChatOpenAI from langchain/chat_models/openaiBaseChatPromptTemplate from langchain/promptsBasePromptTemplate from langchain/promptsSerializedBasePromptTemplate from langchain/promptsrenderTemplate from langchain/promptsAgentAction from langchain/schemaAgentFinish from langchain/schemaAgentStep from langchain/schemaBaseMessage from langchain/schemaHumanMessage from langchain/schemaInputValues from langchain/schemaPartialValues from langchain/schemaSerpAPI from langchain/toolsTool from langchain/toolsCalculator from langchain/tools/calculator
This notebook goes through how to create your own custom agent based on a chat model.
An LLM chat agent consists of three parts:
import { AgentActionOutputParser, AgentExecutor, LLMSingleActionAgent,} from "langchain/agents";import { LLMChain } from "langchain/chains";import { ChatOpenAI } from "langchain/chat_models/openai";import { BaseChatPromptTemplate, BasePromptTemplate, SerializedBasePromptTemplate, renderTemplate,} from "langchain/prompts";import { AgentAction, AgentFinish, AgentStep, BaseMessage, HumanMessage, InputValues, PartialValues,} from "langchain/schema";import { SerpAPI, Tool } from "langchain/tools";import { Calculator } from "langchain/tools/calculator";const PREFIX = `Answer the following questions as best you can. |
4e9727215e95-3171 | You have access to the following tools:`;const formatInstructions = ( toolNames: string) => `Use the following format in your response:Question: the input question you must answerThought: you should always think about what to doAction: the action to take, should be one of [${toolNames}]Action Input: the input to the actionObservation: the result of the action... (this Thought/Action/Action Input/Observation can repeat N times)Thought: I now know the final answerFinal Answer: the final answer to the original input question`;const SUFFIX = `Begin!Question: {input}Thought:{agent_scratchpad}`;class CustomPromptTemplate extends BaseChatPromptTemplate { tools: Tool[]; constructor(args: { tools: Tool[]; inputVariables: string[] }) { super({ inputVariables: args.inputVariables }); this.tools = args.tools; } _getPromptType(): string { throw new |
4e9727215e95-3172 | Error("Not implemented"); } async formatMessages(values: InputValues): Promise<BaseMessage[]> { /** Construct the final template */ const toolStrings = this.tools .map((tool) => `${tool.name}: ${tool.description}`) .join("\n"); const toolNames = this.tools.map((tool) => tool.name).join("\n"); const instructions = formatInstructions(toolNames); const template = [PREFIX, toolStrings, instructions, SUFFIX].join("\n\n"); /** Construct the agent_scratchpad */ const intermediateSteps = values.intermediate_steps as AgentStep[]; const agentScratchpad = intermediateSteps.reduce( (thoughts, { action, observation }) => thoughts + [action.log, `\nObservation: ${observation}`, "Thought:"].join("\n"), "" ); const newInput = { agent_scratchpad: agentScratchpad, ...values }; /** Format the template.
/ const formatted = renderTemplate(template, "f-string", newInput); return [new HumanMessage(formatted)]; } partial(_values: PartialValues): Promise<BaseChatPromptTemplate> { throw new Error("Not implemented"); } serialize(): SerializedBasePromptTemplate { throw new Error("Not implemented"); }}class CustomOutputParser extends AgentActionOutputParser { lc_namespace = ["langchain", "agents", "custom_llm_agent_chat"]; async parse(text: string): Promise<AgentAction | AgentFinish> { if (text.includes("Final Answer:")) { const parts = text.split("Final Answer:"); const input = parts[parts.length - 1].trim(); const finalAnswers = { output: input }; return { log: text, returnValues: finalAnswers }; } const match = /Action: (. *)\nAction Input: (. |
4e9727215e95-3173 | )/s.exec(text); if (!match) { throw new Error(`Could not parse LLM output: ${text}`); } return { tool: match[1].trim(), toolInput: match[2].trim().replace(/^"+|"+$/g, ""), log: text, }; } getFormatInstructions(): string { throw new Error("Not implemented"); }}export const run = async () => { const model = new ChatOpenAI({ temperature: 0 }); const tools = [ new SerpAPI(process.env.SERPAPI_API_KEY, { location: "Austin,Texas,United States", hl: "en", gl: "us", }), new Calculator(), ]; const llmChain = new LLMChain({ prompt: new CustomPromptTemplate({ tools, inputVariables: ["input", "agent_scratchpad"], }), llm: model, }); const agent = new LLMSingleActionAgent({ llmChain, outputParser: new CustomOutputParser(), stop: ["\nObservation"], }); const executor = new AgentExecutor({ agent, tools, }); console.log("Loaded agent. "); const input = `Who is Olivia Wilde's boyfriend? What is his current age raised to the 0.23 power?`; console.log(`Executing with input "${input}"...`); const result = await executor.call({ input }); console.log(`Got output ${result.output}`);};run(); |
4e9727215e95-3174 | API Reference:AgentActionOutputParser from langchain/agentsAgentExecutor from langchain/agentsLLMSingleActionAgent from langchain/agentsLLMChain from langchain/chainsChatOpenAI from langchain/chat_models/openaiBaseChatPromptTemplate from langchain/promptsBasePromptTemplate from langchain/promptsSerializedBasePromptTemplate from langchain/promptsrenderTemplate from langchain/promptsAgentAction from langchain/schemaAgentFinish from langchain/schemaAgentStep from langchain/schemaBaseMessage from langchain/schemaHumanMessage from langchain/schemaInputValues from langchain/schemaPartialValues from langchain/schemaSerpAPI from langchain/toolsTool from langchain/toolsCalculator from langchain/tools/calculator
Logging and tracing
Page Title: Logging and tracing | 🦜️🔗 Langchain
Paragraphs: |
4e9727215e95-3175 | Paragraphs:
Skip to main content🦜️🔗 LangChainDocsUse casesAPILangSmithPython DocsCTRLKGet startedIntroductionInstallationQuickstartModulesModel I/OData connectionChainsMemoryAgentsAgent typesHow-toSubscribing to eventsCancelling requestsCustom LLM AgentCustom LLM Agent (with a ChatModel)Logging and tracingAdding a timeoutToolsToolkitsCallbacksModulesGuidesEcosystemAdditional resourcesCommunity navigatorAPI referenceModulesAgentsHow-toLogging and tracingLogging and tracingYou can pass the verbose flag when creating an agent to enable logging of all events to the console. For example:You can also enable tracing by setting the LANGCHAIN_TRACING environment variable to true.import { initializeAgentExecutorWithOptions } from "langchain/agents";import { OpenAI } from "langchain/llms/openai";import { SerpAPI } from "langchain/tools";import { Calculator } from "langchain/tools/calculator";const model = new OpenAI({ temperature: 0 });const tools = [ new SerpAPI(process.env.SERPAPI_API_KEY, { location: "Austin,Texas,United States", hl: "en", gl: "us", }), new Calculator(),];const executor = await initializeAgentExecutorWithOptions(tools, model, { agentType: "zero-shot-react-description", verbose: true,});const input = `Who is Olivia Wilde's boyfriend? |
4e9727215e95-3176 | What is his current age raised to the 0.23 power?`;const result = await executor.call({ input });API Reference:initializeAgentExecutorWithOptions from langchain/agentsOpenAI from langchain/llms/openaiSerpAPI from langchain/toolsCalculator from langchain/tools/calculator[chain/start] [1:chain:agent_executor] Entering Chain run with input: { "input": "Who is Olivia Wilde's boyfriend? What is his current age raised to the 0.23 power? "}[chain/start] [1:chain:agent_executor > 2:chain:llm_chain] Entering Chain run with input: { "input": "Who is Olivia Wilde's boyfriend? What is his current age raised to the 0.23 power? ", "agent_scratchpad": "", "stop": [ "\nObservation: " ]}[llm/start] [1:chain:agent_executor > 2:chain:llm_chain > 3:llm:openai] Entering LLM run with input: { "prompts": [ "Answer the following questions as best you can. You have access to the following tools:\n\nsearch: a search engine. useful for when you need to answer questions about current events. input should be a search query.\ncalculator: Useful for getting the result of a math expression. |
4e9727215e95-3177 | The input to this tool should be a valid mathematical expression that could be executed by a simple calculator.\n\nUse the following format in your response:\n\nQuestion: the input question you must answer\nThought: you should always think about what to do\nAction: the action to take, should be one of [search,calculator]\nAction Input: the input to the action\nObservation: the result of the action\n... (this Thought/Action/Action Input/Observation can repeat N times)\nThought: I now know the final answer\nFinal Answer: the final answer to the original input question\n\nBegin!\n\nQuestion: Who is Olivia Wilde's boyfriend? |
4e9727215e95-3178 | What is his current age raised to the 0.23 power?\nThought:" ]}[llm/end] [1:chain:agent_executor > 2:chain:llm_chain > 3:llm:openai] [3.52s] Exiting LLM run with output: { "generations": [ [ { "text": " I need to find out who Olivia Wilde's boyfriend is and then calculate his age raised to the 0.23 power.\nAction: search\nAction Input: \"Olivia Wilde boyfriend\"", "generationInfo": { "finishReason": "stop", "logprobs": null } } ] ], "llmOutput": { "tokenUsage": { "completionTokens": 39, "promptTokens": 220, "totalTokens": 259 } }}[chain/end] [1:chain:agent_executor > 2:chain:llm_chain] [3.53s] Exiting Chain run with output: { "text": " I need to find out who Olivia Wilde's boyfriend is and then calculate his age raised to the 0.23 power.\nAction: search\nAction Input: \"Olivia Wilde boyfriend\""}[agent/action] [1:chain:agent_executor] Agent selected action: { "tool": "search", "toolInput": "Olivia Wilde boyfriend", "log": " I need to find out who Olivia Wilde's boyfriend is and then calculate his age raised to the 0.23 power.\nAction: search\nAction |
4e9727215e95-3179 | and then calculate his age raised to the 0.23 power.\nAction: search\nAction Input: \"Olivia Wilde boyfriend\""}[tool/start] [1:chain:agent_executor > 4:tool:search] Entering Tool run with input: "Olivia Wilde boyfriend"[tool/end] [1:chain:agent_executor > 4:tool:search] [845ms] Exiting Tool run with output: "In January 2021, Wilde began dating singer Harry Styles after meeting during the filming of Don't Worry Darling. |
4e9727215e95-3180 | Their relationship ended in November 2022. "[chain/start] [1:chain:agent_executor > 5:chain:llm_chain] Entering Chain run with input: { "input": "Who is Olivia Wilde's boyfriend? What is his current age raised to the 0.23 power? ", "agent_scratchpad": " I need to find out who Olivia Wilde's boyfriend is and then calculate his age raised to the 0.23 power.\nAction: search\nAction Input: \"Olivia Wilde boyfriend\"\nObservation: In January 2021, Wilde began dating singer Harry Styles after meeting during the filming of Don't Worry Darling. Their relationship ended in November 2022.\nThought:", "stop": [ "\nObservation: " ]}[llm/start] [1:chain:agent_executor > 5:chain:llm_chain > 6:llm:openai] Entering LLM run with input: { "prompts": [ "Answer the following questions as best you can. You have access to the following tools:\n\nsearch: a search engine. useful for when you need to answer questions about current events. input should be a search query.\ncalculator: Useful for getting the result of a math expression. |
4e9727215e95-3181 | The input to this tool should be a valid mathematical expression that could be executed by a simple calculator.\n\nUse the following format in your response:\n\nQuestion: the input question you must answer\nThought: you should always think about what to do\nAction: the action to take, should be one of [search,calculator]\nAction Input: the input to the action\nObservation: the result of the action\n... (this Thought/Action/Action Input/Observation can repeat N times)\nThought: I now know the final answer\nFinal Answer: the final answer to the original input question\n\nBegin!\n\nQuestion: Who is Olivia Wilde's boyfriend? What is his current age raised to the 0.23 power?\nThought: I need to find out who Olivia Wilde's boyfriend is and then calculate his age raised to the 0.23 power.\nAction: search\nAction Input: \"Olivia Wilde boyfriend\"\nObservation: In January 2021, Wilde began dating singer Harry Styles after meeting during the filming of Don't Worry Darling. |
4e9727215e95-3182 | Their relationship ended in November 2022.\nThought:" ]}[llm/end] [1:chain:agent_executor > 5:chain:llm_chain > 6:llm:openai] [3.65s] Exiting LLM run with output: { "generations": [ [ { "text": " I need to find out Harry Styles' age.\nAction: search\nAction Input: \"Harry Styles age\"", "generationInfo": { "finishReason": "stop", "logprobs": null } } ] ], "llmOutput": { "tokenUsage": { "completionTokens": 23, "promptTokens": 296, "totalTokens": 319 } }}[chain/end] [1:chain:agent_executor > 5:chain:llm_chain] [3.65s] Exiting Chain run with output: { "text": " I need to find out Harry Styles' age.\nAction: search\nAction Input: \"Harry Styles age\""}[agent/action] [1:chain:agent_executor] Agent selected action: { "tool": "search", "toolInput": "Harry Styles age", "log": " I need to find out Harry Styles' age.\nAction: search\nAction Input: \"Harry Styles age\""}[tool/start] [1:chain:agent_executor > 7:tool:search] Entering Tool run with input: "Harry Styles age"[tool/end] [1:chain:agent_executor > |
4e9727215e95-3183 | run with input: "Harry Styles age"[tool/end] [1:chain:agent_executor > 7:tool:search] [632ms] Exiting Tool run with output: "29 years"[chain/start] [1:chain:agent_executor > 8:chain:llm_chain] Entering Chain run with input: { "input": "Who is Olivia Wilde's boyfriend? What is his current age raised to the 0.23 power? |
4e9727215e95-3184 | ", "agent_scratchpad": " I need to find out who Olivia Wilde's boyfriend is and then calculate his age raised to the 0.23 power.\nAction: search\nAction Input: \"Olivia Wilde boyfriend\"\nObservation: In January 2021, Wilde began dating singer Harry Styles after meeting during the filming of Don't Worry Darling. Their relationship ended in November 2022.\nThought: I need to find out Harry Styles' age.\nAction: search\nAction Input: \"Harry Styles age\"\nObservation: 29 years\nThought:", "stop": [ "\nObservation: " ]}[llm/start] [1:chain:agent_executor > 8:chain:llm_chain > 9:llm:openai] Entering LLM run with input: { "prompts": [ "Answer the following questions as best you can. You have access to the following tools:\n\nsearch: a search engine. useful for when you need to answer questions about current events. input should be a search query.\ncalculator: Useful for getting the result of a math expression. |
4e9727215e95-3185 | The input to this tool should be a valid mathematical expression that could be executed by a simple calculator.\n\nUse the following format in your response:\n\nQuestion: the input question you must answer\nThought: you should always think about what to do\nAction: the action to take, should be one of [search,calculator]\nAction Input: the input to the action\nObservation: the result of the action\n... (this Thought/Action/Action Input/Observation can repeat N times)\nThought: I now know the final answer\nFinal Answer: the final answer to the original input question\n\nBegin!\n\nQuestion: Who is Olivia Wilde's boyfriend? What is his current age raised to the 0.23 power?\nThought: I need to find out who Olivia Wilde's boyfriend is and then calculate his age raised to the 0.23 power.\nAction: search\nAction Input: \"Olivia Wilde boyfriend\"\nObservation: In January 2021, Wilde began dating singer Harry Styles after meeting during the filming of Don't Worry Darling. |
4e9727215e95-3186 | Their relationship ended in November 2022.\nThought: I need to find out Harry Styles' age.\nAction: search\nAction Input: \"Harry Styles age\"\nObservation: 29 years\nThought:" ]}[llm/end] [1:chain:agent_executor > 8:chain:llm_chain > 9:llm:openai] [2.72s] Exiting LLM run with output: { "generations": [ [ { "text": " I need to calculate 29 raised to the 0.23 power.\nAction: calculator\nAction Input: 29^0.23", "generationInfo": { "finishReason": "stop", "logprobs": null } } ] ], "llmOutput": { "tokenUsage": { "completionTokens": 26, "promptTokens": 329, "totalTokens": 355 } }}[chain/end] [1:chain:agent_executor > 8:chain:llm_chain] [2.72s] Exiting Chain run with output: { "text": " I need to calculate 29 raised to the 0.23 power.\nAction: calculator\nAction Input: 29^0.23"}[agent/action] [1:chain:agent_executor] Agent selected action: { "tool": "calculator", "toolInput": "29^0.23", "log": " I need to calculate 29 raised to the 0.23 power.\nAction: calculator\nAction |
4e9727215e95-3187 | need to calculate 29 raised to the 0.23 power.\nAction: calculator\nAction Input: 29^0.23"}[tool/start] [1:chain:agent_executor > 10:tool:calculator] Entering Tool run with input: "29^0.23"[tool/end] [1:chain:agent_executor > 10:tool:calculator] [3ms] Exiting Tool run with output: "2.169459462491557"[chain/start] [1:chain:agent_executor > 11:chain:llm_chain] Entering Chain run with input: { "input": "Who is Olivia Wilde's boyfriend? What is his current age raised to the 0.23 power? |
4e9727215e95-3188 | ", "agent_scratchpad": " I need to find out who Olivia Wilde's boyfriend is and then calculate his age raised to the 0.23 power.\nAction: search\nAction Input: \"Olivia Wilde boyfriend\"\nObservation: In January 2021, Wilde began dating singer Harry Styles after meeting during the filming of Don't Worry Darling. Their relationship ended in November 2022.\nThought: I need to find out Harry Styles' age.\nAction: search\nAction Input: \"Harry Styles age\"\nObservation: 29 years\nThought: I need to calculate 29 raised to the 0.23 power.\nAction: calculator\nAction Input: 29^0.23\nObservation: 2.169459462491557\nThought:", "stop": [ "\nObservation: " ]}[llm/start] [1:chain:agent_executor > 11:chain:llm_chain > 12:llm:openai] Entering LLM run with input: { "prompts": [ "Answer the following questions as best you can. You have access to the following tools:\n\nsearch: a search engine. useful for when you need to answer questions about current events. input should be a search query.\ncalculator: Useful for getting the result of a math expression. |
4e9727215e95-3189 | The input to this tool should be a valid mathematical expression that could be executed by a simple calculator.\n\nUse the following format in your response:\n\nQuestion: the input question you must answer\nThought: you should always think about what to do\nAction: the action to take, should be one of [search,calculator]\nAction Input: the input to the action\nObservation: the result of the action\n... (this Thought/Action/Action Input/Observation can repeat N times)\nThought: I now know the final answer\nFinal Answer: the final answer to the original input question\n\nBegin!\n\nQuestion: Who is Olivia Wilde's boyfriend? What is his current age raised to the 0.23 power?\nThought: I need to find out who Olivia Wilde's boyfriend is and then calculate his age raised to the 0.23 power.\nAction: search\nAction Input: \"Olivia Wilde boyfriend\"\nObservation: In January 2021, Wilde began dating singer Harry Styles after meeting during the filming of Don't Worry Darling. |
4e9727215e95-3190 | Their relationship ended in November 2022.\nThought: I need to find out Harry Styles' age.\nAction: search\nAction Input: \"Harry Styles age\"\nObservation: 29 years\nThought: I need to calculate 29 raised to the 0.23 power.\nAction: calculator\nAction Input: 29^0.23\nObservation: 2.169459462491557\nThought:" ]}[llm/end] [1:chain:agent_executor > 11:chain:llm_chain > 12:llm:openai] [3.51s] Exiting LLM run with output: { "generations": [ [ { "text": " I now know the final answer.\nFinal Answer: Harry Styles is Olivia Wilde's boyfriend and his current age raised to the 0.23 power is 2.169459462491557. ", "generationInfo": { "finishReason": "stop", "logprobs": null } } ] ], "llmOutput": { "tokenUsage": { "completionTokens": 39, "promptTokens": 371, "totalTokens": 410 } }}[chain/end] [1:chain:agent_executor > 11:chain:llm_chain] [3.51s] Exiting Chain run with output: { "text": " I now know the final answer.\nFinal Answer: Harry Styles is Olivia Wilde's boyfriend and his current age raised to the 0.23 power is |
4e9727215e95-3191 | Harry Styles is Olivia Wilde's boyfriend and his current age raised to the 0.23 power is 2.169459462491557. "}[chain/end] [1:chain:agent_executor] [14.90s] Exiting Chain run with output: { "output": "Harry Styles is Olivia Wilde's boyfriend and his current age raised to the 0.23 power is 2.169459462491557. "}PreviousCustom LLM Agent (with a ChatModel)NextAdding a timeoutCommunityDiscordTwitterGitHubPythonJS/TSMoreHomepageBlogCopyright © 2023 LangChain, Inc. |
4e9727215e95-3192 | Get startedIntroductionInstallationQuickstartModulesModel I/OData connectionChainsMemoryAgentsAgent typesHow-toSubscribing to eventsCancelling requestsCustom LLM AgentCustom LLM Agent (with a ChatModel)Logging and tracingAdding a timeoutToolsToolkitsCallbacksModulesGuidesEcosystemAdditional resourcesCommunity navigatorAPI referenceModulesAgentsHow-toLogging and tracingLogging and tracingYou can pass the verbose flag when creating an agent to enable logging of all events to the console. For example:You can also enable tracing by setting the LANGCHAIN_TRACING environment variable to true.import { initializeAgentExecutorWithOptions } from "langchain/agents";import { OpenAI } from "langchain/llms/openai";import { SerpAPI } from "langchain/tools";import { Calculator } from "langchain/tools/calculator";const model = new OpenAI({ temperature: 0 });const tools = [ new SerpAPI(process.env.SERPAPI_API_KEY, { location: "Austin,Texas,United States", hl: "en", gl: "us", }), new Calculator(),];const executor = await initializeAgentExecutorWithOptions(tools, model, { agentType: "zero-shot-react-description", verbose: true,});const input = `Who is Olivia Wilde's boyfriend? |
4e9727215e95-3193 | What is his current age raised to the 0.23 power?`;const result = await executor.call({ input });API Reference:initializeAgentExecutorWithOptions from langchain/agentsOpenAI from langchain/llms/openaiSerpAPI from langchain/toolsCalculator from langchain/tools/calculator[chain/start] [1:chain:agent_executor] Entering Chain run with input: { "input": "Who is Olivia Wilde's boyfriend? What is his current age raised to the 0.23 power? "}[chain/start] [1:chain:agent_executor > 2:chain:llm_chain] Entering Chain run with input: { "input": "Who is Olivia Wilde's boyfriend? What is his current age raised to the 0.23 power? ", "agent_scratchpad": "", "stop": [ "\nObservation: " ]}[llm/start] [1:chain:agent_executor > 2:chain:llm_chain > 3:llm:openai] Entering LLM run with input: { "prompts": [ "Answer the following questions as best you can. You have access to the following tools:\n\nsearch: a search engine. useful for when you need to answer questions about current events. input should be a search query.\ncalculator: Useful for getting the result of a math expression. |
4e9727215e95-3194 | The input to this tool should be a valid mathematical expression that could be executed by a simple calculator.\n\nUse the following format in your response:\n\nQuestion: the input question you must answer\nThought: you should always think about what to do\nAction: the action to take, should be one of [search,calculator]\nAction Input: the input to the action\nObservation: the result of the action\n... (this Thought/Action/Action Input/Observation can repeat N times)\nThought: I now know the final answer\nFinal Answer: the final answer to the original input question\n\nBegin!\n\nQuestion: Who is Olivia Wilde's boyfriend? |
4e9727215e95-3195 | What is his current age raised to the 0.23 power?\nThought:" ]}[llm/end] [1:chain:agent_executor > 2:chain:llm_chain > 3:llm:openai] [3.52s] Exiting LLM run with output: { "generations": [ [ { "text": " I need to find out who Olivia Wilde's boyfriend is and then calculate his age raised to the 0.23 power.\nAction: search\nAction Input: \"Olivia Wilde boyfriend\"", "generationInfo": { "finishReason": "stop", "logprobs": null } } ] ], "llmOutput": { "tokenUsage": { "completionTokens": 39, "promptTokens": 220, "totalTokens": 259 } }}[chain/end] [1:chain:agent_executor > 2:chain:llm_chain] [3.53s] Exiting Chain run with output: { "text": " I need to find out who Olivia Wilde's boyfriend is and then calculate his age raised to the 0.23 power.\nAction: search\nAction Input: \"Olivia Wilde boyfriend\""}[agent/action] [1:chain:agent_executor] Agent selected action: { "tool": "search", "toolInput": "Olivia Wilde boyfriend", "log": " I need to find out who Olivia Wilde's boyfriend is and then calculate his age raised to the 0.23 power.\nAction: search\nAction |
4e9727215e95-3196 | and then calculate his age raised to the 0.23 power.\nAction: search\nAction Input: \"Olivia Wilde boyfriend\""}[tool/start] [1:chain:agent_executor > 4:tool:search] Entering Tool run with input: "Olivia Wilde boyfriend"[tool/end] [1:chain:agent_executor > 4:tool:search] [845ms] Exiting Tool run with output: "In January 2021, Wilde began dating singer Harry Styles after meeting during the filming of Don't Worry Darling. |
4e9727215e95-3197 | Their relationship ended in November 2022. "[chain/start] [1:chain:agent_executor > 5:chain:llm_chain] Entering Chain run with input: { "input": "Who is Olivia Wilde's boyfriend? What is his current age raised to the 0.23 power? ", "agent_scratchpad": " I need to find out who Olivia Wilde's boyfriend is and then calculate his age raised to the 0.23 power.\nAction: search\nAction Input: \"Olivia Wilde boyfriend\"\nObservation: In January 2021, Wilde began dating singer Harry Styles after meeting during the filming of Don't Worry Darling. Their relationship ended in November 2022.\nThought:", "stop": [ "\nObservation: " ]}[llm/start] [1:chain:agent_executor > 5:chain:llm_chain > 6:llm:openai] Entering LLM run with input: { "prompts": [ "Answer the following questions as best you can. You have access to the following tools:\n\nsearch: a search engine. useful for when you need to answer questions about current events. input should be a search query.\ncalculator: Useful for getting the result of a math expression. |
4e9727215e95-3198 | The input to this tool should be a valid mathematical expression that could be executed by a simple calculator.\n\nUse the following format in your response:\n\nQuestion: the input question you must answer\nThought: you should always think about what to do\nAction: the action to take, should be one of [search,calculator]\nAction Input: the input to the action\nObservation: the result of the action\n... (this Thought/Action/Action Input/Observation can repeat N times)\nThought: I now know the final answer\nFinal Answer: the final answer to the original input question\n\nBegin!\n\nQuestion: Who is Olivia Wilde's boyfriend? What is his current age raised to the 0.23 power?\nThought: I need to find out who Olivia Wilde's boyfriend is and then calculate his age raised to the 0.23 power.\nAction: search\nAction Input: \"Olivia Wilde boyfriend\"\nObservation: In January 2021, Wilde began dating singer Harry Styles after meeting during the filming of Don't Worry Darling. |
4e9727215e95-3199 | Their relationship ended in November 2022.\nThought:" ]}[llm/end] [1:chain:agent_executor > 5:chain:llm_chain > 6:llm:openai] [3.65s] Exiting LLM run with output: { "generations": [ [ { "text": " I need to find out Harry Styles' age.\nAction: search\nAction Input: \"Harry Styles age\"", "generationInfo": { "finishReason": "stop", "logprobs": null } } ] ], "llmOutput": { "tokenUsage": { "completionTokens": 23, "promptTokens": 296, "totalTokens": 319 } }}[chain/end] [1:chain:agent_executor > 5:chain:llm_chain] [3.65s] Exiting Chain run with output: { "text": " I need to find out Harry Styles' age.\nAction: search\nAction Input: \"Harry Styles age\""}[agent/action] [1:chain:agent_executor] Agent selected action: { "tool": "search", "toolInput": "Harry Styles age", "log": " I need to find out Harry Styles' age.\nAction: search\nAction Input: \"Harry Styles age\""}[tool/start] [1:chain:agent_executor > 7:tool:search] Entering Tool run with input: "Harry Styles age"[tool/end] [1:chain:agent_executor > |
Subsets and Splits