Update README with dynamic evaluation command and fix potential bug

Author: Just-do-st

README.md

@@ -160,7 +160,7 @@ For more configuration information, please refer to `config.ini` and `.env.examp
 
 Supports dynamic evaluation to verify the secure programming capabilities of Large Language Models (LLMs) in the `instruct` scenario. To start it:
 ```Python3
-uv run -m sec_code_bench.eval_dynamic \
+uv run sec_code_bench/eval_dynamic \
     --benchlanguage java \
     --benchmark ./datasets/runnable/benchmark/java/java.json \
     --llm-under-test "OPENAI::<ModelName>::<APIKey>::<URL>"

README.zh-CN.md

@@ -158,7 +158,7 @@ $ uv run sec_code_bench/eval_static.py --help
 
 支持通过动态运行，验证大语言模型（LLM）在 `instruct` 场景下的安全编程能力。启动方法如下：
 ```Python3
-uv run -m sec_code_bench.eval_dynamic \
+uv run sec_code_bench/eval_dynamic \
     --benchlanguage java \
     --benchmark ./datasets/runnable/benchmark/java/java.json \
     --llm-under-test "OPENAI::<ModelName>::<APIKey>::<URL>"

sec_code_bench/eval_dynamic.py

@@ -215,8 +215,8 @@ async def main():
         return 1
 
     current_dir = os.path.dirname(os.path.abspath(__file__))
-    testcases_paths = [f"{current_dir}/../datasets/runnable/benchmark/{case.get("language")}"
-                       f"/prompts/{case.get("prompt")}.{LOCALE}" for case in data]
+    testcases_paths = [f"{current_dir}/../datasets/runnable/benchmark/{case.get('language')}"
+                       f"/prompts/{case.get('prompt')}.{LOCALE}" for case in data]
 
     testcases_prompts = await get_contents_async(testcases_paths)
 

sec_code_bench/llms/openai.py

@@ -40,6 +40,7 @@ def __init__(self, config: LLMConfig, rate_limiter: AsyncRateLimiter) -> None:
         self.aclient = openai.AsyncOpenAI(
             api_key=self.api_key,
             base_url=self.url,
+            max_retries=0,
         )
         self.rate_limiter = rate_limiter
     @override