openai version and api_key problem

[MarkJiang-maji]

Because openai has been upgraded to v1.9. I have upgraded the askGPT35 function related to openai (inside utils.py) into the newest version, but I still have the problem when running instruction_gen.py. Is this my API key problem? How can I fix this...

my askGPT35 function: def askGPT35(question, model='text-davinci-003', is_response=False, temperature=0.7): sleep_time = 2 api_key = os.environ.get("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") # 這裡確保你的 API 金鑰已經在環境變數中設置 if is_response is True: p_message = [ {'role': 'system', 'content': 'I want you act as a Professional Verilog coder.'}, {'role': 'user', 'content': question} ] else: p_message = [ {'role': 'system', 'content': ''}, {'role': 'user', 'content': question} ] max_gen_tokens = 2048 count = 0 while True: if count == 5: dic = {'finish_reason': 'length', 'text': ''} return [dic] try: client = openai.OpenAI(api_key=api_key) # 初始化 OpenAI 物件，提供 API 金鑰 response = client.chat.completions.create( model=model, messages=p_message, temperature=temperature, max_tokens=max_gen_tokens, ) print(response) ans = response['choices'][0]['message']['content'] dic = {'text': ans, 'finish_reason': response['choices'][0]['finish_reason']} break

    except Exception as e:
        if 'maximum context' in str(e):
            count += 1 
            max_gen_tokens = int(max_gen_tokens / 1.3)
        logging.warning(f"OpenAIError: {e}.")
        logging.warning("Hit request rate limit; retrying...")
        time.sleep(sleep_time)
return [dic]

Terminal: Loaded 10 seed instructions 0%| | 0/50 [00:00<?, ?it/s] WARNING:root:OpenAIError: The api_key client option must be set either by passing api_key to the client or by setting the OPENAI_API_KEY environment variable. WARNING:root:Hit request rate limit; retrying...

[DevinShang]

Hi, could you please set openai api in your main function like this: ''' import openai openai.api_key = 'your api key' ''' and then you do not need to pass the attribute "api_key=api_key" when initializing the client object

[MarkJiang-maji]

Thank you for providing the requirements, but I'm still facing the following issue:

WARNING:root:OpenAIError: Invalid URL (POST /v1/engines/gpt-35-turbo/chat/completions). WARNING:root:Hit request rate limit; retrying...

how can i fix this?

[DevinShang]

We use the Openai service through Azure and there may be slight differences between the two ways. You may change the model name from "gpt-35-turbo" to "gpt-3.5-turbo". After all, if you use the API from the Openai website, we recommend you refer to the official documents for modifying the generation code.