调用kimi-api进行多轮对话

今天练习了调用kimi-api进行多轮对话，测试一下他的文本生成能力

from openai import OpenAI
 
client = OpenAI(
    api_key = "xxxxx",
    base_url = "https://api.moonshot.cn/v1",
)
conversations = [
        {"role": "system", "content": "你是一个乐于解答各种问题的助手，你的任务是为用户提供专业、准确、有见地的建议。"},
 	    {"role": "user", "content": "给我写一篇篮球与鸡的励志故事。"}
]
response = client.chat.completions.create(
    model="moonshot-v1-8k",  # 调用模型
    messages=conversations,
    stream=True,  # 实现流式输出
)
ans = ""
for chunk in response:
    char = chunk.choices[0].delta.content
    if char is not None:
        ans += char
        print(char, end='')  # 打印出返回的回答

# 在聊天记录中增加AI的回复
conversations.append({"role": "assistant", "content": ans}) # 将AI的回复加入到对话中,role为assistant
# 追加问题
next_question = "将上述故事翻译成英文。"
conversations.append({"role": "user", "content": next_question})
response = client.chat.completions.create(
    model="moonshot-v1-8k",  # 调用kimi模型
    messages=conversations,
    stream=True,  # 实现流式输出
)
# 获取后续回答
for chunk in response:
    char = chunk.choices[0].delta.content
    print(char, end='')  # 打印出返回的回答

运行后，流式输出答案~