tirnaltd
commented
10 months ago I've got the same problem. No matter what combo of gpt4 models, max token sizes and temperatures I use, it won't work. I'm approved for GPT4 8k context, and I can use it in the playground. I've noticed in the playground that GPT4 is only available for chats and not completions. Is this relevant to my problem?
let result = try await openAI.sendCompletion(
with: "This is an example text",
model: .other("gpt-4"), // optional OpenAIModelType
maxTokens: 8192, // optional Int?
temperature: 0.1 // optional Double?
)
I'm running the following code using the model .gpt4(.gpt4) but the output is an empty string. My account is approved for gpt-4, which I verified by running gpt-4 in the playground on OpenAI. Any help/ideas is appreciated!
Here is my getResponse code:
public func getResponse(input: String, completion: @escaping (Result<String, Error>) -> Void) { client?.sendCompletion(with: input, model: .gpt4(.gpt4), maxTokens: 500, completionHandler: { result in switch result { case .success(let model): let output = model.choices?.first?.text ?? "" completion(.success(output)) case .failure(let error): completion(.failure(error)) } }) }