Merge 748c1597e9 into a9fc018444

Dockerfile

@@ -0,0 +1,8 @@
+FROM pytorch/pytorch:1.13.1-cuda11.6-cudnn8-runtime
+COPY . .
+RUN pip3  install -r requirements.txt
+ENV model_path="/model"
+
+EXPOSE 7860
+
+CMD [ "python3","web_demo.py" ]

README.md

@@ -133,6 +133,12 @@ curl -X POST "http://127.0.0.1:8000" \
 }
 ```
 
+### 在docker容器中部署
+```bash
+docker build -t chatglm .
+docker run -v /the/model/path:/model chatglm
+```
+
 ## 低成本部署
 ### 模型量化
 默认情况下，模型以 FP16 精度加载，运行上述代码需要大概 13GB 显存。如果你的 GPU 显存有限，可以尝试以量化方式加载模型，使用方法如下：
@@ -156,8 +162,6 @@ model = AutoModel.from_pretrained("THUDM/chatglm-6b-int4", trust_remote_code=Tru
 model = AutoModel.from_pretrained("THUDM/chatglm-6b-int4-qe", trust_remote_code=True).half().cuda()
 ```
 
-
-
 ### CPU 部署
 如果你没有 GPU 硬件的话，也可以在 CPU 上进行推理，但是推理速度会更慢。使用方法如下（需要大概 32GB 内存）
 ```python

web_demo.py

@@ -1,10 +1,14 @@
 from transformers import AutoModel, AutoTokenizer
 import gradio as gr
-
-tokenizer = AutoTokenizer.from_pretrained("THUDM/chatglm-6b", trust_remote_code=True)
-model = AutoModel.from_pretrained("THUDM/chatglm-6b", trust_remote_code=True).half().cuda()
+import os
+modelPath = os.getenv('model_path')
+modelPath = modelPath if  modelPath!="" else "THUDM/chatglm-6b" 
+tokenizer = AutoTokenizer.from_pretrained(modelPath, trust_remote_code=True)
+model = AutoModel.from_pretrained(modelPath, trust_remote_code=True).half().cuda()
 model = model.eval()
 
+
+
 MAX_TURNS = 20
 MAX_BOXES = MAX_TURNS * 2
 
@@ -42,4 +46,4 @@ with gr.Blocks() as demo:
             temperature = gr.Slider(0, 1, value=0.95, step=0.01, label="Temperature", interactive=True)
             button = gr.Button("Generate")
     button.click(predict, [txt, max_length, top_p, temperature, state], [state] + text_boxes)
-demo.queue().launch(share=False, inbrowser=True)
+demo.queue().launch(share=False, inbrowser=True,server_name="0.0.0.0")