diff --git a/README.md b/README.md
index 8bcedb2..972da40 100644
--- a/README.md
+++ b/README.md
@@ -23,6 +23,10 @@
 
 🚩 本项目未涉及微调、训练过程，但可利用微调或训练对本项目效果进行优化。
 
+🐳 Docker镜像：registry.cn-beijing.aliyuncs.com/isafetech/chatmydata:1.0 （感谢 @InkSong🌲 ）
+
+💻 运行方式：docker run -d -p 80:7860 --gpus all registry.cn-beijing.aliyuncs.com/isafetech/chatmydata:1.0 
+
 🌐 [AutoDL 镜像](https://www.codewithgpu.com/i/imClumsyPanda/langchain-ChatGLM/langchain-ChatGLM)
 
 📓 [ModelWhale 在线运行项目](https://www.heywhale.com/mw/project/643977aa446c45f4592a1e59)
@@ -60,6 +64,23 @@
 
     本项目中默认选用的 Embedding 模型 [GanymedeNil/text2vec-large-chinese](https://huggingface.co/GanymedeNil/text2vec-large-chinese/tree/main) 约占用显存 3GB，也可修改为在 CPU 中运行。
 
+## Docker 整合包
+🐳 Docker镜像地址：`registry.cn-beijing.aliyuncs.com/isafetech/chatmydata:1.0 `🌲
+
+💻 一行命令运行：
+```shell
+docker run -d -p 80:7860 --gpus all registry.cn-beijing.aliyuncs.com/isafetech/chatmydata:1.0
+```
+
+- 该版本镜像大小`25.2G`，使用[v0.1.16](https://github.com/imClumsyPanda/langchain-ChatGLM/releases/tag/v0.1.16)，以`nvidia/cuda:12.1.1-cudnn8-runtime-ubuntu22.04`为基础镜像
+- 该版本内置两个`embedding`模型：`m3e-base`，`text2vec-large-chinese`，内置`fastchat+chatglm-6b`
+- 该版本目标为方便一键部署使用，请确保您已经在Linux发行版上安装了NVIDIA驱动程序
+- 请注意，您不需要在主机系统上安装CUDA工具包，但需要安装`NVIDIA Driver`以及`NVIDIA Container Toolkit`，请参考[安装指南](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html)
+- 首次拉取和启动均需要一定时间，首次启动时请参照下图使用`docker logs -f <container id>`查看日志
+- 如遇到启动过程卡在`Waiting..`步骤，建议使用`docker exec -it <container id> bash`进入`/logs/`目录查看对应阶段日志
+![](img/docker_logs.png)
+
+
 ## Docker 部署
 为了能让容器使用主机GPU资源，需要在主机上安装 [NVIDIA Container Toolkit](https://github.com/NVIDIA/nvidia-container-toolkit)。具体安装步骤如下：
 ```shell
@@ -230,7 +251,7 @@ Web UI 可以实现如下功能：
 - [x] VUE 前端
 
 ## 项目交流群
-<img src="img/qr_code_36.jpg" alt="二维码" width="300" height="300" />
+<img src="img/qr_code_39.jpg" alt="二维码" width="300" height="300" />
 
 
 🎉 langchain-ChatGLM 项目微信交流群，如果你也对本项目感兴趣，欢迎加入群聊参与讨论交流。
diff --git a/api.py b/api.py
index 4965cbf..cf947b0 100644
--- a/api.py
+++ b/api.py
@@ -4,7 +4,7 @@ import os
 import shutil
 from typing import List, Optional
 import urllib
-
+import asyncio
 import nltk
 import pydantic
 import uvicorn
@@ -382,6 +382,7 @@ async def stream_chat(websocket: WebSocket):
         for resp, history in local_doc_qa.get_knowledge_based_answer(
                 query=question, vs_path=vs_path, chat_history=history, streaming=True
         ):
+            await asyncio.sleep(0)
             await websocket.send_text(resp["result"][last_print_len:])
             last_print_len = len(resp["result"])
 
diff --git a/img/docker_logs.png b/img/docker_logs.png
new file mode 100644
index 0000000..0382958
Binary files /dev/null and b/img/docker_logs.png differ
diff --git a/img/qr_code_39.jpg b/img/qr_code_39.jpg
new file mode 100644
index 0000000..e866af0
Binary files /dev/null and b/img/qr_code_39.jpg differ
diff --git a/models/loader/loader.py b/models/loader/loader.py
index f315e6c..0c32835 100644
--- a/models/loader/loader.py
+++ b/models/loader/loader.py
@@ -257,10 +257,21 @@ class LoaderCheckPoint:
         # 在调用chat或者stream_chat时,input_ids会被放到model.device上
         # 如果transformer.word_embeddings.device和model.device不同,则会导致RuntimeError
         # 因此这里将transformer.word_embeddings,transformer.final_layernorm,lm_head都放到第一张卡上
-        device_map = {f'{layer_prefix}.word_embeddings': 0,
+        
+        encode = ""
+        if 'chatglm2' in self.model_name:
+            device_map = {
+                f"{layer_prefix}.embedding.word_embeddings": 0,
+                f"{layer_prefix}.rotary_pos_emb": 0,
+                f"{layer_prefix}.output_layer": 0,
+                f"{layer_prefix}.encoder.final_layernorm": 0,
+                f"base_model.model.output_layer": 0 
+            }
+            encode = ".encoder"
+        else:
+            device_map = {f'{layer_prefix}.word_embeddings': 0,
                       f'{layer_prefix}.final_layernorm': 0, 'lm_head': 0,
                       f'base_model.model.lm_head': 0, }
-
         used = 2
         gpu_target = 0
         for i in range(num_trans_layers):
@@ -268,7 +279,7 @@ class LoaderCheckPoint:
                 gpu_target += 1
                 used = 0
             assert gpu_target < num_gpus
-            device_map[f'{layer_prefix}.layers.{i}'] = gpu_target
+            device_map[f'{layer_prefix}{encode}.layers.{i}'] = gpu_target
             used += 1
 
         return device_map
diff --git a/webui_st.py b/webui_st.py
index 6d1265e..bd12a3a 100644
--- a/webui_st.py
+++ b/webui_st.py
@@ -143,7 +143,7 @@ def init_model(llm_model: str = 'chat-glm-6b', embedding_model: str = 'text2vec'
 #     return history + [[None, model_status]]
 
 
-def get_vector_store(vs_id, files, sentence_size, history, one_conent, one_content_segmentation):
+def get_vector_store(local_doc_qa, vs_id, files, sentence_size, history, one_conent, one_content_segmentation):
     vs_path = os.path.join(KB_ROOT_PATH, vs_id, "vector_store")
     filelist = []
     if not os.path.exists(os.path.join(KB_ROOT_PATH, vs_id, "content")):
@@ -455,6 +455,8 @@ with st.sidebar:
             cols = st.columns([12, 10])
             kb_name = cols[0].text_input(
                 '新知识库名称', placeholder='新知识库名称', label_visibility='collapsed')
+            if 'kb_name' not in st.session_state:
+                st.session_state.kb_name = kb_name
             cols[1].button('新建知识库', on_click=on_new_kb)
             vs_path = st.selectbox(
                 '选择知识库', vs_list, on_change=on_vs_change, key='vs_path')