tensorchord/modelz-llm

bug: Completion request returns wrong response

Closed this issue · 0 comments

https://platform.openai.com/docs/api-reference/completions/create

It should return:

{
  "id": "cmpl-uqkvlQyYK7bGYrRHQ0eXlWi7",
  "object": "text_completion",
  "created": 1589478378,
  "model": "text-davinci-003",
  "choices": [
    {
      "text": "\n\nThis is indeed a test",
      "index": 0,
      "logprobs": null,
      "finish_reason": "length"
    }
  ],
  "usage": {
    "prompt_tokens": 5,
    "completion_tokens": 7,
    "total_tokens": 12
  }
}

But we get:

{
    "id": "1306c25e-968f-4a02-adcb-5bcfbd19524a",
    "object": "chat",
    "created": "2023-06-20T19:58:12.437203",
    "model": "bigscience/bloomz-560m",
    "usage": {
        "prompt_tokens": 7,
        "completion_tokens": 5,
        "total_tokens": 12
    },
    "choices": [
        {
            "message": {
                "content": " The Dark Knight Rises",
                "role": "assistant",
                "name": ""
            },
            "index": 0,
            "finish_reason": "stop"
        }
    ]
}