llama-server: Where can I find detailed explanations for the timing sub-items in the chat completions response?

[dev-jonghoonpark]

hello.

Where can I find detailed explanations for the timing sub-items in the chat completions response?

• prompt_n
• prompt_ms
• prompt_per_token_ms
• prompt_per_second
• predicted_n
• predicted_ms
• predicted_per_token_ms
• predicted_per_second

I couldn't find them at https://github.com/ggml-org/llama.cpp/tree/master/tools/server, and they don't seem to appear in search results under that repository either.

If they don't exist, could they be added?
While I can roughly infer their meanings, having a clear explanation documented would be helpful.

Thanks.