update chatglm deploy

13-model_deploy_guide/container_deploy/Dockerfile

@@ -0,0 +1,2 @@
+FROM nginx
+RUN echo '这是一个本地构建的nginx镜像' > /usr/share/nginx/html/index.html

13-model_deploy_guide/container_deploy/README.md

@@ -1,6 +1,8 @@
 # 镜像与容器 概念介绍
 **docker：管理镜像和容器**
+
 ![docker](https://www.runoob.com/wp-content/uploads/2016/04/docker01.png)
+
 [docker 入门教程](https://www.runoob.com/docker/docker-tutorial.html)
 
 # 理解docker
@@ -12,12 +14,12 @@
 - 镜像是可以保存的，就可以不用每次都run dockerfile了。
 
 # 一些重要网站
-[docker 官网](https://www.docker.com)
-[docker github](https://github.com/docker/docker-ce)
-[docker guide page](https://dockerdocs.cn/get-started/index.html)
-[dockerfile]https://www.runoob.com/docker/docker-dockerfile.html
-[docker 命令查询](https://docs.docker.com/engine/reference/commandline/docker/)
-[dockerfile 命令查询](https://docs.docker.com/engine/reference/builder/#understand-how-cmd-and-entrypoint-interact)
+[docker 官网](https://www.docker.com)<br>
+[docker github](https://github.com/docker/docker-ce)<br>
+[docker guide page](https://dockerdocs.cn/get-started/index.html)<br>
+[dockerfile]https://www.runoob.com/docker/docker-dockerfile.html<br>
+[docker 命令查询](https://docs.docker.com/engine/reference/commandline/docker/)<br>
+[dockerfile 命令查询](https://docs.docker.com/engine/reference/builder/#understand-how-cmd-and-entrypoint-interact)<br>
 
 #得到image的几种方法
 方法1：docker pull    # 拉取仓库的镜像
@@ -90,4 +92,4 @@ EXPOSE 5000
 # 启动应用程序
 CMD ["python", "app.py"]
 ```
-**docker build . 从上述dockerfile中创建镜像**
+**docker build -t . 从上述dockerfile中创建镜像**

13-model_deploy_guide/cpp_deploy/mnist/CMakeLists.txt

@@ -3,10 +3,11 @@ project(mnist)
 set(CMAKE_CXX_STANDARD 14)
 
 # 设置自己的 torch 路径
-set(Torch_DIR /home/elvin/.conda/envs/pytorch_env/lib/python3.8/site-packages/torch/share/cmake/Torch)
+#set(Torch_DIR /home/elvin/.conda/envs/pytorch_env/lib/python3.8/site-packages/torch/share/cmake/Torch)
+set(Torch_DIR /home/elvin/miniconda3/envs/python3.10/lib/python3.10/site-packages/torch/share/cmake/Torch)
 find_package(Torch REQUIRED)
 
-option(DOWNLOAD_MNIST "Download the MNIST dataset from the internet" ON)
+option(DOWNLOAD_MNIST "Download the MNIST dataset from the internet" OFF)
 if (DOWNLOAD_MNIST)
   message(STATUS "Downloading MNIST dataset")
   execute_process(

13-model_deploy_guide/cpp_deploy/mnist/mnist-infer.cpp

@@ -1,11 +1,12 @@
-#include <torch/torch.h>
+#include <torch/torch.h> // 
 #include <torch/script.h>
 #include <cstddef>
 #include <cstdio>
 #include <iostream>
 #include <string>
 #include <vector>
-void  mnist_train() {
+
+int  mnist_train() {
   torch::DeviceType device_type = torch::kCPU;
   torch::Device device(device_type);
   auto input = torch::rand({1, 1, 28, 28}, torch::kFloat32);
@@ -14,11 +15,11 @@ void  mnist_train() {
   auto max_result = output.max(1, true);
   auto max_index = std::get<1>(max_result).item<float>();
   std::cout << "=========cpu max_index: " << max_index << std::endl;
+  return max_index;
 }
 
 int main() {
   mnist_train();
   std::cout << "run mnist-infer.cpp successfully !!!" << std::endl;
   return 0;
 }
-

13-model_deploy_guide/flask_deploy/flask_demo/hello.py

@@ -0,0 +1,10 @@
+from flask import Flask
+
+app = Flask(__name__)
+
+@app.route('/')
+def hello_world():
+   return 'Hello, World!'
+
+if __name__ =="__main__":
+  app.run(host='0.0.0.0', port=5000)

13-model_deploy_guide/flask_deploy/flask_demo/resnet_infer.py

@@ -1,5 +1,5 @@
 import onnx
-import onnxruntime
+#import onnxruntime
 import numpy as np
 from PIL import Image
 import torch
@@ -55,4 +55,4 @@ def image_infer(path, model_path):
 if __name__ == "__main__":
     get_onnx_model("resnet50.onnx")
     # image_infer("cat.jpg", "resnet50.onnx")
-    print("run resnet_infer.py successfully")
+    print("run resnet_infer.py successfully")

13-model_deploy_guide/flask_deploy/flask_demo/upload_image.py

@@ -60,4 +60,5 @@ def upload():
     # app.run(host='0.0.0.0', port=8987, debug=True)
     app.run(host='0.0.0.0', port=8987)
     # app.run()
+
 

13-model_deploy_guide/java_deploy/README.md

@@ -0,0 +1,31 @@
+# 1. 部署我们自己的聊天机器人
+
+## 1.1 代码准备
+git clone [email protected]:wenda-LLM/wenda.git 
+*自己download 也可以*
+
+## 1.2 模型准备
+- [链接](https://pan.baidu.com/s/1VPRGReHfnnqe_ULKjquoaQ?pwd=oaae)
+- 提取码：oaae 
+- 将pretrained model 放到 wenda/model 下（建议）；
+
+## 1.3 环境准备
+```shell
+- conda create -n python3.10 python=3.10（建议）
+- pip install torch==1.13.1+cu116 torchvision==0.14.1+cu116 torchaudio==0.13.1 --extra-index-url https://download.pytorch.org/whl/cu116
+- cd */wenda/requirements/ 
+- pip install -r requirements.txt # update transformers==4.27.1
+- pip install requirements-glm6b-lora.txt
+- pip install protobuf transformers==4.27.1 cpm_kernels
+```
+## 1.4 运行程序
+cp example.config.yml config.yml
+source run_GLM6B.sh
+
+## 1.5 连接网络进行访问
+http://172.29.240.181:17860
+**注意：IP要换成自己的IP，端口号 和 config.yml 保存一致**
+
+# 2 [参考资料]
+[参考资料2](https://huggingface.co/THUDM/chatglm-6b-int4)
+[参考资料](https://github.com/wenda-LLM/wenda/tree/main)

14-model_learning/chatGLM/README.md → 15-model_learning/chatGLM/README.md

@@ -1,6 +1,13 @@
 # 聊天机器人
+
+## 1.1 环境搭建
 ```shell
-pip install protobuf transformers==4.27.1 cpm_kernels
+- conda create -n python3.10 python=3.10（建议）
+- pip install torch==1.13.1+cu116 torchvision==0.14.1+cu116 torchaudio==0.13.1 --extra-index-url https://download.pytorch.org/whl/cu116
+- cd */wenda/requirements/ 
+- pip install -r requirements.txt # update transformers==4.27.1
+- pip install requirements-glm6b-lora.txt
+- pip install protobuf transformers==4.27.1 cpm_kernels
 ```
 ## coding
 ```python

6-save_load_guide/save_load_demo.py

@@ -5,30 +5,30 @@
 import torch.nn.functional as F
 import torch.optim as optim
 
-# class Net(nn.Module):
-#     def __init__(self):
-#         super(Net, self).__init__()
-#         self.conv1 = nn.Conv2d(1, 32, 3, 1)
-#         self.conv2 = nn.Conv2d(32, 64, 3, 1)
-#         self.dropout1 = nn.Dropout(0.25)
-#         self.dropout2 = nn.Dropout(0.5)
-#         self.fc1 = nn.Linear(9216, 128)
-#         self.fc2 = nn.Linear(128, 10)
+class Net(nn.Module):
+    def __init__(self):
+        super(Net, self).__init__()
+        self.conv1 = nn.Conv2d(1, 32, 3, 1)
+        self.conv2 = nn.Conv2d(32, 64, 3, 1)
+        self.dropout1 = nn.Dropout(0.25)
+        self.dropout2 = nn.Dropout(0.5)
+        self.fc1 = nn.Linear(9216, 128)
+        self.fc2 = nn.Linear(128, 10)
 
-#     def forward(self, x):
-#         x = self.conv1(x)
-#         x = F.relu(x)
-#         x = self.conv2(x)
-#         x = F.relu(x)
-#         x = F.max_pool2d(x, 2)
-#         x = self.dropout1(x)
-#         x = torch.flatten(x, 1)
-#         x = self.fc1(x)
-#         x = F.relu(x)
-#         x = self.dropout2(x)
-#         x = self.fc2(x)
-#         output = F.log_softmax(x, dim=1)
-#         return output
+    def forward(self, x):
+        x = self.conv1(x)
+        x = F.relu(x)
+        x = self.conv2(x)
+        x = F.relu(x)
+        x = F.max_pool2d(x, 2)
+        x = self.dropout1(x)
+        x = torch.flatten(x, 1)
+        x = self.fc1(x)
+        x = F.relu(x)
+        x = self.dropout2(x)
+        x = self.fc2(x)
+        output = F.log_softmax(x, dim=1)
+        return output
 
 def save_demo_v1():
     model = Net()
@@ -93,7 +93,7 @@ def load_ckpt_demo():
     print("output shape: ", output.shape)
 
 def save_trace_model():
-    model = Net()
+    model = Net().eval()
     # 通过trace 得到了一个新的model，我们最终保存的是这个新的model
     traced_model = torch.jit.trace(model, torch.randn(1, 1, 28, 28))
     traced_model.save("traced_model.pt")
@@ -112,8 +112,8 @@ def load_trace_model():
     # load_para_demo()
     # tensor_save()
     # load_to_gpu()
-    # save_trace_model()
-    load_trace_model()
+    save_trace_model()
+    # load_trace_model()
     # save_ckpt_demo()
     # load_ckpt_demo()
-    print("run save_load_demo.py successfully !!!")
+    print("run save_load_demo.py successfully !!!")