| 30 | |
| 31 | |
| 32 | def inference_chat(chat, model, api_url, token): |
| 33 | |
| 34 | messages = [] |
| 35 | for role, content in chat: |
| 36 | messages.append({"role": role, "content": content}) |
| 37 | |
| 38 | client = OpenAI( |
| 39 | # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx", |
| 40 | api_key=token, |
| 41 | base_url=api_url, |
| 42 | ) |
| 43 | |
| 44 | |
| 45 | num_try = 5 |
| 46 | for _ in range(num_try): |
| 47 | try: |
| 48 | completion = client.chat.completions.create( |
| 49 | model=model, # 此处以qwen-plus为例，可按需更换模型名称。模型列表：https://help.aliyun.com/zh/model-studio/getting-started/models |
| 50 | messages=messages |
| 51 | ) |
| 52 | except: |
| 53 | print("Network Error:") |
| 54 | try: |
| 55 | print(completion.model_dump_json()) |
| 56 | except: |
| 57 | print("Request Failed") |
| 58 | time.sleep(2) |
| 59 | else: |
| 60 | break |
| 61 | |
| 62 | |
| 63 | return json.loads(completion.model_dump_json())['choices'][0]['message']['content'] |
| 64 | |
| 65 | # headers = { |
| 66 | # "Content-Type": "application/json", |
| 67 | # "Authorization": f"Bearer {token}" |
| 68 | # } |
| 69 | |
| 70 | # data = { |
| 71 | # "model": model, |
| 72 | # "messages": [], |
| 73 | # "max_tokens": 2048, |
| 74 | # 'temperature': 0.0, |
| 75 | # "seed": 1234 |
| 76 | # } |
| 77 | |