llama-server why does the legacy completion response use .content vs .choices[0].text

[codefromthecrypt]

Hi, I am trying to understand the version of the /v1/completions endpoint llama-server is emulating, and I can't seem to find it. When I make a completion request, the content is returned in a top-level field .content, where as far as I can tell, openai wants it as .choices[0].text. Any insight on this?

from llama-server

curl -s -X POST localhost:8080/v1/completions -H "Content-Type: application/json" -d '{
>   "model": "qwen2:0.5b",
>   "prompt": "<|im_start|>user\nWhich ocean contains the falkland islands?\n<|im_end|>\n<|im_start|>assistant\n"
> }' |jq .
{
  "content": "The Falkland Islands are located in the South Atlantic Ocean.",
--snip--

from ollama

$ curl -s -X POST localhost:11434/v1/completions -H "Content-Type: application/json" -d '{
>   "model": "qwen2:0.5b",
>   "prompt": "<|im_start|>user\nWhich ocean contains the falkland islands?\n<|im_end|>\n<|im_start|>assistant\n"
> }' | jq .
{
  "id": "cmpl-843",
  "object": "text_completion",
  "created": 1724843529,
  "model": "qwen2:0.5b",
  "system_fingerprint": "fp_ollama",
  "choices": [
    {
      "text": "The Falklands Islands are part of the Atlantic Ocean.",
      "index": 0,
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 25,
    "completion_tokens": 12,
    "total_tokens": 37
  }
}