LangChain AgentExecutor call()抱怨需要一个输入键,但在传递输入时得到2个。抱怨删除时缺少输入

我正在使用LangChain Typescript(0.0.96)构建一个可以回答问题的简单聊天机器人。我使用的是AgentExecutor,它使用从DB和EntityMemory中查找信息的工具作为内存。以下是代码的样子:

// Create memory input
const input = {"input": question};
// Load the relevant history using the entities used in the question and the conversation
relevantEntityHistory = await this.memory.loadMemoryVariables(input);

console.log('Relevant Entity History', relevantEntityHistory);
/**
* Relevant Entity History {
*  history: '',
*  entities: {
*    HTML: 'No current information known.',
*    AI: 'No current information known.',
*    JavaScript: 'No current information known.'
*  }
*}
*/

// Create the agent executor
if (!this.executor) {
   // Build the agent executor with options
   this.executor = await initializeAgentExecutorWithOptions(this.tools, this.model, {
       agentType: "chat-conversational-react-description",
       verbose: true,
       memory: this.memory,
    });
}

console.log("Executing with input: ", question);
// Executing with input:  Please create an HTML page with three columns. The left column will have a list of conversations the customer had with our AI chatbot. The middle column will contain all messages for a selected conversation and a text area for customers to send new messages. The right column will have a list of knowledge base sources that the chatbot can use to answer customer questions.

const result = await this.executor.call({ 
    input: question,
    chat_history: relevantEntityHistory.history,
});

问题似乎出在 输入:问题 属性发送到executor调用方法。上面的代码给了我这个错误:

错误:需要一个输入键,但得到2个

我尝试删除输入,有趣的是,它似乎能够开始执行链,但后来它抱怨“输入变量缺少值 input "

[chain/start] [1:chain:AgentExecutor] Entering Chain run with input: {
  "chat_history": [],
  "history": "",
  "entities": {
    "Human: \n\nOutput: NONE": "No current information known."
  }
}

[chain/start] [1:chain:AgentExecutor > 2:chain:LLMChain] Entering Chain run with input: {
  "chat_history": [],
  "history": "",
  "entities": {
    "Human: \n\nOutput: NONE": "No current information known."
  },
  "agent_scratchpad": [],
  "stop": [
    "Observation:"
  ]
}

[chain/error] [1:chain:AgentExecutor > 2:chain:LLMChain] [3ms] Chain run errored with error: "Missing value for input variable `input`"

[chain/error] [1:chain:AgentExecutor] [8ms] Chain run errored with error: "Missing value for input variable `input`"

    at C:\Users\Sheri\autodev\node_modules\langchain\dist\prompts\chat.cjs:224:27
    at Array.reduce (<anonymous>)
    at ChatPromptTemplate.formatMessages (C:\Users\Sheri\autodev\node_modules\langchain\dist\prompts\chat.cjs:222:62)
    at process.processTicksAndRejections (node:internal/process/task_queues:95:5)
    at async ChatPromptTemplate.formatPromptValue (C:\Users\Sheri\autodev\node_modules\langchain\dist\prompts\chat.cjs:115:32)
    at async LLMChain._call (C:\Users\Sheri\autodev\node_modules\langchain\dist\chains\llm_chain.cjs:120:29)
    at async LLMChain.call (C:\Users\Sheri\autodev\node_modules\langchain\dist\chains\base.cjs:80:29)
    at async LLMChain.predict (C:\Users\Sheri\autodev\node_modules\langchain\dist\chains\llm_chain.cjs:139:24)
    at async ChatConversationalAgent._plan (C:\Users\Sheri\autodev\node_modules\langchain\dist\agents\agent.cjs:211:24)
    at async AgentExecutor._call (C:\Users\Sheri\autodev\node_modules\langchain\dist\agents\executor.cjs:88:28)

任何关于我遗漏了什么或如何在我的情况下正确传递“输入”的指针都将不胜感激。谢谢

快速更新

我尝试使用executor run()方法而不是call(),尽管我不完全清楚它们之间的区别(Typescript文档仍然有点单薄)。以下是代码的这一部分:

const result = await this.executor.run({
    input: question,
});

但我还是或多或少地收到了同样的错误

Error retrieving response from  as knowledge base using OpenAI: Error: Chain agent_executor expects multiple inputs, cannot use 'run'
    at AgentExecutor.run (C:\Users\Sheri\autodev\node_modules\langchain\dist\chains\base.cjs:53:19)
    at WebsiteBasedConversationAgent.<anonymous> (C:\Users\Sheri\autodev\dist\services\agents\WebsiteBasedConversationAgent.js:99:48)
    at Generator.next (<anonymous>)
    at fulfilled (C:\Users\Sheri\autodev\dist\services\agents\WebsiteBasedConversationAgent.js:5:58)
    at process.processTicksAndRejections (node:internal/process/task_queues:95:5)

我还试着在代码中注释loadMemoryVariables()行,以防它以某种方式导致重复输入但没有骰子