From 657889e3e4c0fa41676f29ed4b147f1a88e07424 Mon Sep 17 00:00:00 2001
From: Shaojun Liu <61072813+liu-shaojun@users.noreply.github.com>
Date: Tue, 24 Sep 2024 17:40:50 +0800
Subject: [PATCH] use english prompt by default (#12115)

---
 .../serving/xpu/docker/vllm_online_benchmark.py    | 14 ++++++++------
 1 file changed, 8 insertions(+), 6 deletions(-)

diff --git a/docker/llm/serving/xpu/docker/vllm_online_benchmark.py b/docker/llm/serving/xpu/docker/vllm_online_benchmark.py
index f852f955a7e..8d4e30954ea 100644
--- a/docker/llm/serving/xpu/docker/vllm_online_benchmark.py
+++ b/docker/llm/serving/xpu/docker/vllm_online_benchmark.py
@@ -435,12 +435,14 @@ def benchmark(llm_urls, model, prompt, num_requests, max_concurrent_requests, ma
 MODEL = "/llm/models/" + model_name
 MAX_TOKENS = output_length  # 修改 MAX_TOKENS 为 output_length
 
-if "Qwen" not in MODEL and "chatglm" not in MODEL:
-    # print("using Llama PROMPT")
-    PROMPT = ENGLISH_PROMPT
-else:
-    # print("using Qwen/chatglm PROMPT")
-    PROMPT = CHINESE_PROMPT
+# if "Qwen" not in MODEL and "chatglm" not in MODEL:
+#     print("using Llama PROMPT")
+#     PROMPT = ENGLISH_PROMPT
+# else:
+#     print("using Qwen/chatglm PROMPT")
+#     PROMPT = CHINESE_PROMPT
+
+PROMPT = ENGLISH_PROMPT
 
 # 加载模型的 tokenizer
 from transformers import AutoTokenizer