rewrite web_demo.py with gradio.Chatbot and support load model from local directory

README.md

@@ -53,6 +53,8 @@ ChatGLM-6B 使用了和 ChatGPT 相似的技术，针对中文问答和对话进
 ```
 完整的模型实现可以在 [Hugging Face Hub](https://huggingface.co/THUDM/chatglm-6b) 上查看。如果你从 Hugging Face Hub 上下载checkpoint的速度较慢，也可以从[这里](https://cloud.tsinghua.edu.cn/d/fb9f16d6dc8f482596c2/)手动下载。
 
+手动下载的模型放置在 `model` 目录下
+
 ### Demo
 
 我们提供了一个基于 [Gradio](https://gradio.app) 的网页版 Demo 和一个命令行 Demo。使用时首先需要下载本仓库：

README_en.md

@@ -51,6 +51,8 @@ Generate dialogue with the following code
 
 The full model implementation is on [HuggingFace Hub](https://huggingface.co/THUDM/chatglm-6b).
 
+Or you can download model manually and put at `model` directory.
+
 ### Demo
 
 We provide a Web demo based on [Gradio](https://gradio.app) and a command line demo in the repo. First clone our repo with:

cli_demo.py

@@ -2,8 +2,10 @@ import os
 import platform
 from transformers import AutoTokenizer, AutoModel
 
-tokenizer = AutoTokenizer.from_pretrained("THUDM/chatglm-6b", trust_remote_code=True)
-model = AutoModel.from_pretrained("THUDM/chatglm-6b", trust_remote_code=True).half().cuda()
+MODEL_ID = "./model" if os.path.exists('./model') else "THUDM/chatglm-6b"
+
+tokenizer = AutoTokenizer.from_pretrained(MODEL_ID, trust_remote_code=True)
+model = AutoModel.from_pretrained(MODEL_ID, trust_remote_code=True).half().cuda()
 model = model.eval()
 
 os_name = platform.system()

web_demo.py

@@ -1,45 +1,77 @@
+import os
 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()
+MODEL_ID = "./model" if os.path.exists('./model') else "THUDM/chatglm-6b"
+
+tokenizer = AutoTokenizer.from_pretrained(MODEL_ID, trust_remote_code=True)
+model = AutoModel.from_pretrained(MODEL_ID, trust_remote_code=True).half().cuda()
 model = model.eval()
 
 MAX_TURNS = 20
-MAX_BOXES = MAX_TURNS * 2
+
+WELCOME_PROMPT = [[None, "[ChatGLM-6B]：Welcome, please input text and press enter"]]
 
 
-def predict(input, max_length, top_p, temperature, history=None):
-    if history is None:
-        history = []
-    for response, history in model.stream_chat(tokenizer, input, history, max_length=max_length, top_p=top_p,
-                                               temperature=temperature):
-        updates = []
+def predict(input, max_length, top_p, temperature, history):
+    for _, history in model.stream_chat(
+            tokenizer, input, history,
+            max_length=max_length,
+            top_p=top_p,
+            temperature=temperature,
+    ):
+        chatbot = []
+
         for query, response in history:
-            updates.append(gr.update(visible=True, value="用户：" + query))
-            updates.append(gr.update(visible=True, value="ChatGLM-6B：" + response))
-        if len(updates) < MAX_BOXES:
-            updates = updates + [gr.Textbox.update(visible=False)] * (MAX_BOXES - len(updates))
-        yield [history] + updates
+            chatbot.append([
+                "[用户]：" + query,
+                "[ChatGLM-6B]：" + response
+            ])
+
+        if len(chatbot) > MAX_TURNS:
+            chatbot = chatbot[- MAX_TURNS:]
+
+        yield history, WELCOME_PROMPT + chatbot
 
 
-with gr.Blocks() as demo:
-    state = gr.State([])
-    text_boxes = []
-    for i in range(MAX_BOXES):
-        if i % 2 == 0:
-            text_boxes.append(gr.Markdown(visible=False, label="提问："))
-        else:
-            text_boxes.append(gr.Markdown(visible=False, label="回复："))
+with gr.Blocks(title="ChatGLM-6B", css='#main-chatbot { height: 480px; }') as demo:
+    input_cache = gr.State()
+    history = gr.State([])
 
     with gr.Row():
-        with gr.Column(scale=4):
-            txt = gr.Textbox(show_label=False, placeholder="Enter text and press enter", lines=11).style(
-                container=False)
-        with gr.Column(scale=1):
-            max_length = gr.Slider(0, 4096, value=2048, step=1.0, label="Maximum length", interactive=True)
-            top_p = gr.Slider(0, 1, value=0.7, step=0.01, label="Top P", interactive=True)
-            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)
+        with gr.Column():
+            pass
+        with gr.Column():
+            chatbot = gr.Chatbot(
+                show_label=False,
+                elem_id="main-chatbot"
+            )
+            input = gr.Textbox(
+                show_label=False,
+                placeholder="Input text and press enter",
+                interactive=True,
+            )
+            with gr.Box():
+                max_length = gr.Slider(0, 4096, value=2048, step=1.0, label="Maximum length", interactive=True)
+                top_p = gr.Slider(0, 1, value=0.7, step=0.01, label="Top P", interactive=True)
+                temperature = gr.Slider(0, 1, value=0.95, step=0.01, label="Temperature", interactive=True)
+        with gr.Column():
+            pass
+
+    input.submit(
+        lambda x: ("", x),
+        [input],
+        [input, input_cache]
+    ).then(
+        predict,
+        [input_cache, max_length, top_p, temperature, history],
+        [history, chatbot],
+    )
+
+    demo.load(
+        lambda: WELCOME_PROMPT,
+        None,
+        [chatbot]
+    )
+
 demo.queue().launch(share=False, inbrowser=True)