From 0772884b06fc9c81db8188fdae3bdfe083cfac54 Mon Sep 17 00:00:00 2001
From: Pierrick HYMBERT <pierrick.hymbert@gmail.com>
Date: Tue, 20 Feb 2024 22:55:29 +0100
Subject: [PATCH] server: tests: add a constant seed in completion request

---
 examples/server/tests/features/server.feature | 2 +-
 examples/server/tests/features/steps/steps.py | 9 ++++++---
 2 files changed, 7 insertions(+), 4 deletions(-)

diff --git a/examples/server/tests/features/server.feature b/examples/server/tests/features/server.feature
index 35b4244d3..968e288d0 100644
--- a/examples/server/tests/features/server.feature
+++ b/examples/server/tests/features/server.feature
@@ -1,7 +1,7 @@
 Feature: llama.cpp server
 
   Background: Server startup
-    Given a server listening on localhost:8080 with 2 slots
+    Given a server listening on localhost:8080 with 2 slots and 42 as seed
     Then  the server is starting
     Then  the server is healthy
 
diff --git a/examples/server/tests/features/steps/steps.py b/examples/server/tests/features/steps/steps.py
index 72857c45e..400b3c126 100644
--- a/examples/server/tests/features/steps/steps.py
+++ b/examples/server/tests/features/steps/steps.py
@@ -7,11 +7,12 @@ import requests
 from behave import step
 
 
-@step(u"a server listening on {server_fqdn}:{server_port} with {n_slots} slots")
-def step_server_config(context, server_fqdn, server_port, n_slots):
+@step(u"a server listening on {server_fqdn}:{server_port} with {n_slots} slots and {seed} as seed")
+def step_server_config(context, server_fqdn, server_port, n_slots, seed):
     context.server_fqdn = server_fqdn
     context.server_port = int(server_port)
     context.n_slots = int(n_slots)
+    context.seed = int(seed)
     context.base_url = f'http://{context.server_fqdn}:{context.server_port}'
 
     context.completions = []
@@ -154,6 +155,7 @@ def request_completion(context, prompt, n_predict=None):
     response = requests.post(f'{context.base_url}/completion', json={
         "prompt": prompt,
         "n_predict": int(n_predict) if n_predict is not None else 4096,
+        "seed": context.seed
     })
     assert response.status_code == 200
     context.completions.append(response.json())
@@ -173,7 +175,8 @@ def oai_chat_completions(context, user_prompt):
         ],
         model=context.model,
         max_tokens=context.max_tokens,
-        stream=context.enable_streaming
+        stream=context.enable_streaming,
+        seed = context.seed
     )
     if context.enable_streaming:
         completion_response = {