diff --git a/README.md b/README.md
index ae6802b..3636420 100644
--- a/README.md
+++ b/README.md
@@ -107,6 +107,9 @@
 <details>
     <summary>Click to expand</summary>
 
+- 2023-08-29 v0.0.8 update:
+    - Fixed missing `embedding_extract`
+    - Fixed default parameters of LLM
 - 2023-08-11 v0.0.7 update:
   - Optimize layout, remove the plugin option, and put the extract vector model option on the home page.
   - The tips are translated into English for easy communication.
diff --git a/docs/README_zh.md b/docs/README_zh.md
index c580e18..eaa1978 100644
--- a/docs/README_zh.md
+++ b/docs/README_zh.md
@@ -101,6 +101,9 @@
 ```
 
 #### 更新日志
+- 2023-08-29 v0.0.8 update:
+  - 修复找不到`embedding_extract`
+  - 修复LLM的默认参数问题
 - 2023-08-11 v0.0.7 update:
   - 优化布局，去掉插件选项，将提取向量模型选项放到主页部分
   - 将提示语英语化，便于交流使用。
diff --git a/knowledge_qa_llm/config.yaml b/knowledge_qa_llm/config.yaml
index 7541a5e..12b318d 100644
--- a/knowledge_qa_llm/config.yaml
+++ b/knowledge_qa_llm/config.yaml
@@ -1,5 +1,5 @@
 title: 🧐 Knowledge QA LLM
-version: 0.0.7
+version: 0.0.8
 
 LLM_API:
   Qwen7B_Chat: your_api
diff --git a/webui.py b/webui.py
index dc83651..31562f9 100644
--- a/webui.py
+++ b/webui.py
@@ -44,7 +44,7 @@ def init_sidebar():
         "top_p",
         min_value=param_top.get("min_value"),
         max_value=param_top.get("max_value"),
-        value=param_top.get("value"),
+        value=param_top.get("default"),
         step=param_top.get("step"),
         help=param_top.get("tip"),
     )
@@ -55,7 +55,7 @@ def init_sidebar():
         "temperature",
         min_value=param_temp.get("min_value"),
         max_value=param_temp.get("max_value"),
-        value=param_temp.get("value"),
+        value=param_temp.get("default"),
         step=param_temp.get("stemp"),
         help=param_temp.get("tip"),
     )
@@ -213,9 +213,6 @@ def tips(txt: str, wait_time: int = 2, icon: str = "🎉"):
     db_path = config.get("vector_db_path")
     db_tools = DBUtils(db_path)
 
-    init_sidebar()
-    init_state()
-
     llm_module = importlib.import_module("knowledge_qa_llm.llm")
     MODEL_OPTIONS = {
         name: getattr(llm_module, name)(api)
@@ -243,6 +240,9 @@ def tips(txt: str, wait_time: int = 2, icon: str = "🎉"):
 
     embedding_extract = init_encoder(ENCODER_OPTIONS[select_encoder])
 
+    init_sidebar()
+    init_state()
+
     input_prompt_container = st.container()
     with input_prompt_container:
         with st.expander("💡Prompt", expanded=False):