From 1cc0155d04918cb3017afa472acea51b77483c4a Mon Sep 17 00:00:00 2001 From: Georgi Gerganov Date: Mon, 20 May 2024 10:16:41 +0300 Subject: [PATCH] server : tuning tests (#7388) * server : don't pass temperature as string * server : increase timeout * tests : fix the fix 0.8f -> 0.8 ggml-ci * tests : set explicit temperature --- examples/server/tests/features/results.feature | 2 ++ examples/server/tests/features/steps/steps.py | 4 ++-- 2 files changed, 4 insertions(+), 2 deletions(-) diff --git a/examples/server/tests/features/results.feature b/examples/server/tests/features/results.feature index 5deb278c2..4ab8ad20c 100644 --- a/examples/server/tests/features/results.feature +++ b/examples/server/tests/features/results.feature @@ -13,6 +13,7 @@ Feature: Results Scenario Outline: consistent results with same seed Given slots + And 0.0 temperature Then the server is starting Then the server is healthy @@ -30,6 +31,7 @@ Feature: Results Scenario Outline: different results with different seed Given slots + And 1.0 temperature 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 7da503f2c..26d9359d7 100644 --- a/examples/server/tests/features/steps/steps.py +++ b/examples/server/tests/features/steps/steps.py @@ -199,7 +199,7 @@ async def step_wait_for_the_server_to_be_started(context, expecting_status): case 'ready' | 'idle': await wait_for_health_status(context, context.base_url, 200, 'ok', - timeout=10, + timeout=30, params={'fail_on_no_slot': 0, 'include_slots': 0}, slots_idle=context.n_slots, slots_processing=0, @@ -883,7 +883,7 @@ async def request_completion(prompt, "cache_prompt": cache_prompt, "id_slot": id_slot, "seed": seed if seed is not None else 42, - "temperature": temperature if temperature is not None else "0.8f", + "temperature": temperature if temperature is not None else 0.8, "n_probs": 2, }, headers=headers,