OpenAI module giving error 400 json: cannot unmarshal array into Go struct field CompletionRequest.prompt of type string\n

[djaffer]

Checked other resources

• [X] I added a very descriptive title to this issue.
• [X] I searched the LangChain.js documentation with the integrated search.
• [X] I used the GitHub search to find a similar question and didn't find it.
• [X] I am sure that this is a bug in LangChain.js rather than my code.
• [X] The bug is not resolved by updating to the latest stable version of LangChain (or the specific integration package).

Example Code

const llm = new OpenAI({
  model: model,
   apiKey: process.env.ASKCYPH_API_KEY,
   configuration: {
      baseURL: customurl,
  },
  verbose: true,
 });
const model = await llm.invoke('what is 2 + 2');

Tested with OpenAI library direct and it works perfectly fine.

Error Message and Stack Trace (if applicable)

} [llm/error] [1:llm:OpenAI] [270ms] LLM run errored with error: "400 json: cannot unmarshal array into Go struct field CompletionRequest.prompt of type string\n\nError: 400 json: cannot unmarshal array into Go struct field CompletionRequest.prompt of type string\n at APIError.generate (/home/langchain/node_modules/openai/error.js:45:20)\n at OpenAI.makeStatusError (/home/langchain/node_modules/openai/core.js:275:33)\n at OpenAI.makeRequest (/home/langchain/node_modules/openai/core.js:318:30)\n at process.processTicksAndRejections (node:internal/process/task_queues:95:5)\n at async /home/langchain/node_modules/@langchain/openai/dist/llms.cjs:517:29\n at async RetryOperation._fn (/home/langchain/node_modules/p-retry/index.js:50:12)" node:internal/process/promises:289 triggerUncaughtException(err, true / fromPromise /); ^

BadRequestError: 400 json: cannot unmarshal array into Go struct field CompletionRequest.prompt of type string at APIError.generate (/home/langchain/node_modules/openai/error.js:45:20) at OpenAI.makeStatusError (/home/langchain/node_modules/openai/core.js:275:33) at OpenAI.makeRequest (/home/langchain/node_modules/openai/core.js:318:30) at process.processTicksAndRejections (node:internal/process/task_queues:95:5) at async /home/langchain/node_modules/@langchain/openai/dist/llms.cjs:517:29 at async RetryOperation._fn (/home/langchain/node_modules/p-retry/index.js:50:12) { status: 400, headers: { }, request_id: undefined, error: { message: 'json: cannot unmarshal array into Go struct field CompletionRequest.prompt of type string', type: 'invalid_request_error', param: null, code: null }, code: null, param: null, type: 'invalid_request_error', attemptNumber: 1, retriesLeft: 6 }

Description

Error in response when using OpenAI library and not ChatOpenAI.

const llm = new OpenAI({
  model: model,
   apiKey: process.env.ASKCYPH_API_KEY,
   configuration: {
      baseURL: url,
  },
  verbose: true,
 });
const model = await llm.invoke(inputText);

System Info

"@langchain/community": "^0.2.25",
"@langchain/openai": "^0.2.6",
"langchain": "^0.2.13",

[dosubot[bot]]

To resolve the 400 error related to unmarshalling a JSON array into a Go struct field CompletionRequest.prompt of type string when using the OpenAI module in LangChain.js, ensure that the prompts parameter passed to the _generate method is an array of strings. The error occurs because the CompletionRequest.prompt field in the Go struct expects a string, but it is receiving an array.

Here is the relevant part of the code where the prompts parameter is used:

async _generate(
  prompts: string[],
  options: this["ParsedCallOptions"],
  runManager?: CallbackManagerForLLMRun
): Promise<LLMResult> {
  const requestStartTime = Date.now();
  const generations = await super._generate(prompts, options, runManager);
  for (let i = 0; i < generations.generations.length; i += 1) {
    const requestEndTime = Date.now();
    const parsedResp = {
      text: generations.generations[i][0].text,
      llm_output: generations.llmOutput,
    };

    const promptLayerRespBody = await promptLayerTrackRequest(
      this.caller,
      "langchain.PromptLayerOpenAI",
      { ...this._identifyingParams(), prompt: prompts[i] } as any,
      this.plTags,
      parsedResp,
      requestStartTime,
      requestEndTime,
      this.promptLayerApiKey
    );

    let promptLayerRequestId;
    if (this.returnPromptLayerId === true) {
      if (promptLayerRespBody && promptLayerRespBody.success === true) {
        promptLayerRequestId = promptLayerRespBody.request_id;
      }

      generations.generations[i][0].generationInfo = {
        ...generations.generations[i][0].generationInfo,
        promptLayerRequestId,
      };
    }
  }

  return generations;
}

Ensure that the prompts parameter is correctly formatted as an array of strings when calling the _generate method. For example:

const prompts = ["Your first prompt", "Your second prompt"];
const options = {}; // Your options here
const runManager = undefined; // Or your run manager instance

const result = await promptLayerOpenAIInstance._generate(prompts, options, runManager);

By ensuring that prompts is an array of strings, you can avoid the 400 error related to unmarshalling a JSON array into a Go struct field CompletionRequest.prompt of type string ^[1].

To continue talking to Dosu, mention @dosu.