server : add system_fingerprint to chat/completion (#10917)

* server : add system_fingerprint to chat/completion

* update README

examples/server/README.md

@@ -724,7 +724,8 @@ This endpoint is public (no API key check). By default, it is read-only. To make
   },
   "total_slots": 1,
   "model_path": "../models/Meta-Llama-3.1-8B-Instruct-Q4_K_M.gguf",
-  "chat_template": "..."
+  "chat_template": "...",
+  "build_info": "b(build number)-(build commit hash)"
 }
 ```
 

examples/server/server.cpp

@@ -598,6 +598,7 @@ struct server_task_result_cmpl_final : server_task_result {
             {"choices",            json::array({choice})},
             {"created",            t},
             {"model",              oaicompat_model},
+            {"system_fingerprint", build_info},
             {"object",             "chat.completion"},
             {"usage", json {
                 {"completion_tokens", n_decoded},
@@ -636,6 +637,7 @@ struct server_task_result_cmpl_final : server_task_result {
             {"created",            t},
             {"id",                 oaicompat_cmpl_id},
             {"model",              oaicompat_model},
+            {"system_fingerprint", build_info},
             {"object",             "chat.completion.chunk"},
             {"usage", json {
                 {"completion_tokens", n_decoded},
@@ -765,6 +767,7 @@ struct server_task_result_cmpl_partial : server_task_result {
             {"created",            t},
             {"id",                 oaicompat_cmpl_id},
             {"model",              oaicompat_model},
+            {"system_fingerprint", build_info},
             {"object",             "chat.completion.chunk"}
         };
 
@@ -3476,6 +3479,7 @@ int main(int argc, char ** argv) {
             { "total_slots",                 ctx_server.params_base.n_parallel },
             { "model_path",                  ctx_server.params_base.model },
             { "chat_template",               llama_get_chat_template(ctx_server.model) },
+            { "build_info",                  build_info },
         };
 
         res_ok(res, data);

examples/server/tests/unit/test_chat_completion.py

@@ -31,6 +31,7 @@ def test_chat_completion(model, system_prompt, user_prompt, max_tokens, re_conte
     })
     assert res.status_code == 200
     assert "cmpl" in res.body["id"] # make sure the completion id has the expected format
+    assert res.body["system_fingerprint"].startswith("b")
     assert res.body["model"] == model if model is not None else server.model_alias
     assert res.body["usage"]["prompt_tokens"] == n_prompt
     assert res.body["usage"]["completion_tokens"] == n_predicted
@@ -63,6 +64,7 @@ def test_chat_completion_stream(system_prompt, user_prompt, max_tokens, re_conte
     last_cmpl_id = None
     for data in res:
         choice = data["choices"][0]
+        assert data["system_fingerprint"].startswith("b")
         assert "gpt-3.5" in data["model"] # DEFAULT_OAICOMPAT_MODEL, maybe changed in the future
         if last_cmpl_id is None:
             last_cmpl_id = data["id"]
@@ -92,6 +94,7 @@ def test_chat_completion_with_openai_library():
         seed=42,
         temperature=0.8,
     )
+    assert res.system_fingerprint is not None and res.system_fingerprint.startswith("b")
     assert res.choices[0].finish_reason == "length"
     assert res.choices[0].message.content is not None
     assert match_regex("(Suddenly)+", res.choices[0].message.content)

examples/server/utils.hpp

@@ -56,6 +56,8 @@ static T json_value(const json & body, const std::string & key, const T & defaul
     }
 }
 
+const static std::string build_info("b" + std::to_string(LLAMA_BUILD_NUMBER) + "-" + LLAMA_COMMIT);
+
 //
 // tokenizer and input processing utils
 //