1

.gitignore

@@ -1 +1,2 @@
 demo/flagged/log.csv
+*.pyc

demo/chatbot/agent.py

@@ -0,0 +1,8 @@
+
+def agent_framework():
+    pass
+
+def agent_train():
+    pass
+def agent_main():
+    pass

demo/chatbot/app.py

@@ -0,0 +1,40 @@
+from fastapi import FastAPI, Request
+import uvicorn
+from datetime import datetime, time, date
+import gradio as gr
+
+from .agent import agent_framework, agent_main
+CUSTOM_PATH = "/agent_bot"
+app = FastAPI()
+
+@app.get("/index")
+def root():
+    return {"message": "hello world!!!"}
+
+with gr.Blocks() as demo:
+    chatbot = gr.Chatbot(show_label=False, height=600, avatar_images=["https://hong.greatdk.com/_next/image?url=%2Fself.jpg&w=1920&q=75", "https://hong.greatdk.com/_next/image?url=%2Fnpc.jpg&w=1920&q=75"])
+    with gr.Row() as input_row:
+        msg_textbox = gr.Textbox(show_label=False, scale=15)
+        submit_btn = gr.Button(value="发送", scale=1)
+
+    clear_btn = gr.ClearButton([msg_textbox, chatbot],value="清除会话")
+    def respond(message, chat_history):
+        print(f"message:{message}, strpped:{message.strip()}")
+        print(f"history:{chat_history}")
+        if len(message.strip()) <= 0 :
+            print(f"message is empty, ignore the message")
+            return "", chat_history
+        
+        init_agent = agent_main.AgentMain()
+        af = agent_framework.AgentFramework(init_agent=init_agent)
+        bot_message = af.run(message.strip(), chat_history)
+        chat_history.append((message.strip(), bot_message))
+        return "", chat_history
+
+    msg_textbox.submit(respond, [msg_textbox, chatbot], [msg_textbox, chatbot]) # 在输入框按回车也进行发送
+    submit_btn.click(respond, [msg_textbox, chatbot], [msg_textbox, chatbot]) # 绑定按钮点击函数
+    
+fastapp = gr.mount_gradio_app(app, demo, path=CUSTOM_PATH)
+
+if __name__=='__main__':
+    uvicorn.run(fastapp, host="0.0.0.0", port=8000)

demo/demo1.py

@@ -1,7 +1,17 @@
+#!/usr/bin/env python
+# -*- encoding: utf-8 -*-
+"""
+@Contact :   liuyuqi.gov@msn.cn
+@Time    :   2024/06/04
+@License :   Copyright © 2017-2022 liuyuqi. All Rights Reserved.
+@Desc    :   实现将图片变为灰度图
+"""
+
 import gradio as gr
 import cv2
 
-def to_black(image):
+def to_black(image: gr.Image)->gr.Image:
+    """ 图片变为灰度 """
     output = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
     return output
 
@@ -9,4 +19,7 @@ def to_black(image):
 interface = gr.Interface(fn=to_black, inputs="image", outputs="image",  examples=[["test.png"]])
 
 if __name__=='__main__':
-    app,local_url, share_url= interface.launch(inbrowser=True, server_name='0.0.0.0')
+    app,local_url, share_url= interface.launch(inbrowser=True, server_name='0.0.0.0',share=True,
+                                               auth=("admin", "123456"), server_port=8000, show_error=True,
+                                               inbrowser=True, inline=False, validate=False,
+                                               title="图片变为灰度图", description="图片变为灰度图")

demo/demo4/app.py

@@ -0,0 +1,51 @@
+import typing
+
+from fastapi import FastAPI
+from pydantic import BaseModel
+import gradio as gr
+
+from demo4.predict import predict
+
+app = FastAPI()
+
+class Request(BaseModel):
+    question: str
+
+class Result(BaseModel):
+    score: float
+    title: str
+    text: str
+
+
+class Response(BaseModel):
+    results: typing.List[Result]
+
+
+@app.post("/predict", response_model=Response)
+async def predict_api(request: Request):
+    results = predict(request.question)
+    return Response(
+        results=[
+            Result(score=r["score"], title=r["title"], text=r["text"]) for r in results
+        ]
+    )
+
+
+def gradio_predict(question: str):
+    results = predict(question)
+    best_result = results[0]
+    return f"{best_result['title']}\n\n{best_result['text']}", best_result["score"]
+
+
+demo = gr.Interface(
+    fn=gradio_predict,
+    inputs=gr.Textbox(
+        label="Ask a question", placeholder="What is the capital of France?"
+    ),
+    outputs=[gr.Textbox(label="Answer"), gr.Number(label="Score")],
+    allow_flagging="never",
+)
+
+# app, local_url, share_url=demo.launch()
+
+fastapp = gr.mount_gradio_app(app, demo, path="/")

demo/demo4/predict.py

@@ -0,0 +1,26 @@
+"""
+This examples demonstrates the setup for Question-Answer-Retrieval.
+
+You can input a query or a question. The script then uses semantic search
+to find relevant passages in Simple English Wikipedia (as it is smaller and fits better in RAM).
+
+As model, we use: nq-distilbert-base-v1
+
+It was trained on the Natural Questions dataset, a dataset with real questions from Google Search
+together with annotated data from Wikipedia providing the answer. For the passages, we encode the
+Wikipedia article tile together with the individual text passages.
+
+Google Colab Example: https://colab.research.google.com/drive/11GunvCqJuebfeTlgbJWkIMT0xJH6PWF1?usp=sharing
+"""
+import json
+import time
+import gzip
+import os
+
+base_directory = os.path.dirname(os.path.realpath(__file__))
+
+def predict(query: str):
+    pass
+
+if __name__ == "__main__":
+    predict("What is the capital of Germany?")

demo/demo5.py

@@ -0,0 +1,40 @@
+#!/usr/bin/env python
+# -*- encoding: utf-8 -*-
+"""
+@Contact :   liuyuqi.gov@msn.cn
+@Time    :   2024/06/04
+@License :   Copyright © 2017-2022 liuyuqi. All Rights Reserved.
+@Desc    :   
+"""
+
+import numpy as np
+import tensorflow as tf
+from deepexplain.tensorflow import DeepExplain
+from tensorflow.keras import backend as K
+from tensorflow.keras.models import Sequential, Model
+import gradio as gr
+
+model = tf.keras.applications.MobileNet()
+
+def saliency(model, x, y):
+    y = y.reshape(1, 1, 1, 1000)
+    with DeepExplain(session=K.get_session()) as de:
+        input_tensor = model.layers[0].input
+        fModel = Model(inputs=input_tensor, outputs = model.layers[-3].output)
+        target_tensor = fModel(input_tensor)
+
+        attributions_gradin = de.explain('grad*input', target_tensor, input_tensor, x, ys=y)
+        sal = np.sum(np.abs(attributions_gradin.squeeze()), axis=-1)
+        sal = (sal - sal.min()) / (sal.max() - sal.min())
+        return sal
+
+inp = gr.inputs.ImageUpload()
+out = gr.outputs.Label(label_names='imagenet1000', max_label_words=1, word_delimiter=",")
+
+io = gr.Interface(inputs=inp,
+                      outputs=out,
+                      model=model,
+                      model_type='keras',
+                      saliency=saliency)
+
+io.launch();

demo/image_classification/app.py

@@ -0,0 +1,29 @@
+#!/usr/bin/env python
+# -*- encoding: utf-8 -*-
+"""
+@Contact :   liuyuqi.gov@msn.cn
+@Time    :   2024/06/04
+@License :   Copyright © 2017-2022 liuyuqi. All Rights Reserved.
+@Desc    :   文本分类
+"""
+import gradio as gr 
+import pandas as pd 
+from ultralytics import YOLO
+from skimage import data
+from PIL import Image
+ 
+model = YOLO('yolov8n-cls.pt')
+
+def predict(img):
+    result = model.predict(source=img)
+    df = pd.Series(result[0].names).to_frame()
+    df.columns = ['names']
+    df['probs'] = result[0].probs
+    df = df.sort_values('probs',ascending=False)
+    res = dict(zip(df['names'],df['probs']))
+    return res
+
+gr.close_all() 
+demo = gr.Interface(fn = predict,inputs = gr.Image(type='pil'), outputs = gr.Label(num_top_classes=5), 
+                    examples = ['cat.jpeg','people.jpeg','coffee.jpeg'])
+demo.launch()

demo/object_detection/app.py

@@ -0,0 +1,40 @@
+import gradio as gr 
+import pandas as pd 
+from skimage import data
+from ultralytics.yolo.data import utils 
+ 
+model = YOLO('yolov8n.pt')
+ 
+#load class_names
+yaml_path = str(Path(ultralytics.__file__).parent/'datasets/coco128.yaml') 
+class_names = utils.yaml_load(yaml_path)['names']
+
+def detect(img):
+    if isinstance(img,str):
+        img = get_url_img(img) if img.startswith('http') else Image.open(img).convert('RGB')
+    result = model.predict(source=img)
+    if len(result[0].boxes.boxes)>0:
+        vis = plots.plot_detection(img,boxes=result[0].boxes.boxes,
+                     class_names=class_names, min_score=0.2)
+    else:
+        vis = img
+    return vis
+    
+with gr.Blocks() as demo:
+    gr.Markdown("# yolov8目标检测演示")
+ 
+    with gr.Tab("捕捉摄像头喔"):
+        in_img = gr.Image(source='webcam',type='pil')
+        button = gr.Button("执行检测",variant="primary")
+ 
+        gr.Markdown("## 预测输出")
+        out_img = gr.Image(type='pil')
+ 
+        button.click(detect,
+                     inputs=in_img, 
+                     outputs=out_img)
+        
+    
+gr.close_all() 
+demo.queue(concurrency_count=5)
+demo.launch()

demo/reading_comprehension/app.py

@@ -0,0 +1,22 @@
+import gradio as gr
+from transformers import pipeline
+
+
+title = "抽取式问答"
+#标题下的描述，支持md格式
+description = "输入上下文与问题后，点击submit按钮，可从上下文中抽取出答案，赶快试试吧！"
+#输入样例
+examples = [
+    ["普希金从那里学习人民的语言，吸取了许多有益的养料，这一切对普希金后来的创作产生了很大的影响。这两年里，普希金创作了不少优秀的作品，如《囚徒》、《致大海》、《致凯恩》和《假如生活欺骗了你》等几十首抒情诗，叙事诗《努林伯爵》，历史剧《鲍里斯·戈都诺夫》，以及《叶甫盖尼·奥涅金》前六章。", "著名诗歌《假如生活欺骗了你》的作者是"],
+    ["普希金从那里学习人民的语言，吸取了许多有益的养料，这一切对普希金后来的创作产生了很大的影响。这两年里，普希金创作了不少优秀的作品，如《囚徒》、《致大海》、《致凯恩》和《假如生活欺骗了你》等几十首抒情诗，叙事诗《努林伯爵》，历史剧《鲍里斯·戈都诺夫》，以及《叶甫盖尼·奥涅金》前六章。", "普希金创作的叙事诗叫什么"]
+    ]
+article = "感兴趣的小伙伴可以阅读[gradio专栏](https://xx)"
+
+
+if __name__=='__main__':
+    gr.Interface.from_pipeline(
+        pipeline("question-answering", model="uer/roberta-base-chinese-extractive-qa"),
+        title=title, description=description, examples=examples, article=article).launch()
+
+
+

demo/test_chatbot.py

@@ -0,0 +1,14 @@
+#!/usr/bin/env python
+# -*- encoding: utf-8 -*-
+"""
+@Contact :   liuyuqi.gov@msn.cn
+@Time    :   2024/06/04
+@License :   Copyright © 2017-2022 liuyuqi. All Rights Reserved.
+@Desc    :   
+"""
+
+from chatbot.app import fastapp
+
+if __name__=='__main__':
+    import uvicorn
+    uvicorn.run(fastapp, host='0.0.0.0', port=8000)

demo/test_demo4.py

@@ -0,0 +1,5 @@
+from demo4.app import fastapp
+
+if __name__=='__main__':
+    import uvicorn
+    uvicorn.run(fastapp, host="0.0.0.0", port=8000)

demo/test_pygwalker.py

@@ -0,0 +1,26 @@
+#!/usr/bin/env python
+# -*- encoding: utf-8 -*-
+"""
+@Contact :   liuyuqi.gov@msn.cn
+@Time    :   2024/06/04
+@License :   Copyright © 2017-2022 liuyuqi. All Rights Reserved.
+@Desc    :   pygwalker 作图
+"""
+
+from ast import main
+import pygwalker as pw
+import gradio as gr
+from datasets imoprt load_dataset
+
+dataset = load_dataset("gradio/NYC-Airbnb-Open-Data", split="train")
+df = dataset.to_pandas()
+
+
+with gr.Blocks() as demo:
+    gr.Label("在 PyGWalker 和 Gradio 中可视化探索 NYC-Airbnb 数据")
+    gr.Markdown("这是一个使用 pygwalker 和 gradio 库构建的数据应用程序。您可以使用拖放操作来探索数据。现在开始您的分析吧！")
+    gr.HTML(pw.walk(dataset=df, spec="./viz-config.json", debug=False, return_html=True))
+
+
+if __name__=='__main__':
+    demo.launch()

demo/text_classification/app.py

@@ -0,0 +1,21 @@
+#!/usr/bin/env python
+# -*- encoding: utf-8 -*-
+"""
+@Contact :   liuyuqi.gov@msn.cn
+@Time    :   2024/06/04
+@License :   Copyright © 2017-2022 liuyuqi. All Rights Reserved.
+@Desc    :   文本分类
+"""
+
+from pyexpat import model
+import gradio as gr
+import pip
+from transformers import pipeline
+
+model = pipeline("zero-shot-classification", model="facebook/bart-large-mnli")
+
+if __name__=='__main__':
+    gr.Interface.from_pipeline(
+        pipeline_name="zero-shot-classification",
+        pipeline=model,
+    ).launch(server_port=8000)

docs/_sidebar.md

@@ -5,4 +5,7 @@
   * [输入组件](input_components.md)
   * [web 开发](web.md)
 * 部署
-  * [部署到云](deploy.md)
+  * [部署到云](deploy.md)
+  * [实例](examples.md)
+
+  

docs/deploy.md

@@ -6,7 +6,16 @@
 
 机器学习项目投入生产环境，技术栈：flask, gunicorn, nginx，gradio, fastapi,docker 等
 
+```
+docker build -t wiki-qa .
+docker run -p 8000:8000 wiki-qa
+
+```
+
+
+## huggingface托管
 
+spaces创建一个项目，安装 transformers, 将代码添加到 Spaces
 
 
 ## 桌面应用打包

docs/examples.md

@@ -0,0 +1,47 @@
+# examples
+
+## resnet18 图片分类
+
+
+## tts 文本转语音
+
+## 聊天机器人
+
+
+## 文本分类
+
+text_classification
+
+## 阅读理解
+
+reading_comprehension
+
+## 机器翻译
+
+machine_translation
+
+
+## 目标检测
+
+object_detection
+
+## 人脸检测
+
+## AI写诗
+
+## AI写代码
+
+## AI写小说
+
+
+## AI生成图片 stable diffusion
+
+
+
+## AI生成代码再生成图片
+
+
+
+## PyGWalker + gradio
+
+pygwalker 是一个将数据转化为交互式可视化应用程序（类似 Tableau 或 PowerBI）的 Python 库

docs/input_components.md

@@ -2,6 +2,172 @@
 
 丰富组件。
 
+* 应用界面：gr.Interface(简易场景)， gr.Blocks(定制化场景)
+* 输入输出：gr.Image(图像)，gr.Audio(音频) , gr.Textbox(文本框)， gr.DataFrame(数据框)， gr.* Dropdown(下拉选项),，gr.Number(数字)， gr.Markdown, gr.Files, gr.Slider(滑动条)， gr.Checkbox(选择框)， gr.Radio(单选框)， gr.MultiCheckbox(多选框)， gr.MultiDropdown(多选下拉框),gr.Button(按钮)
+* 布局组件：gr.Tab(标签页)， gr.Row(行布局),，gr.Column(列布局)
+
+raise gr.Error("Cannot divide by zero!")
+
+
+## Interface
+
+Interface 默认布局左右两栏，函数有三个参数：处理函数，输入，输出
+
+```
+def greet(name):
+    return "Hello, " + name + "!"
+demo = gr.Interface(fn=greet, inputs="text", outputs="text")
+```
+
+以上表示，用户输入文本，经过 greet 处理后，输出文本。此外可以输入文字，图片，选项框等
+
+
+```
+
+demo = gr.Interface(
+    fn=greet,
+    inputs=gr.Textbox(lines=2, placeholder="Name Here...", label="Input Your Name"),
+    outputs="text",
+)
+```
+
+以上输入是一个多行的文本框。
+
+
+```
+def greet(name, is_morning, temperature):
+    salutation = "Good morning" if is_morning else "Good evening"
+    greeting = f"{salutation} {name}. It is {temperature} degrees today"
+    celsius = (temperature - 32) * 5 / 9
+    return greeting, round(celsius, 2)
+
+demo = gr.Interface(
+    fn=greet,
+    inputs=["text", "checkbox", gr.Slider(0, 100)],
+    outputs=["text", "number"],
+)
+```
+以上输入三个表单，界面将在左边显示文本框，选择按钮，滑动按钮供用户操作。同时greet函数就需要定义三个输入变量，两个输出变量
+
+
+iface = gr.Interface(
+    greet,
+    ["number", gr.inputs.Radio(["add", "subtract", "multiply", "divide"]), "number", gr.Image(source="webcam", streaming=True)],
+    "number",
+    live=True,
+)
+
+live=True设置流模式,图片设置来源于摄像头
+
+
+## Blocks
+
+相比 Interface 而言，Blocks可定制更强。
+
+
+```
+with gr.Blocks() as demo:
+    with gr.Row():
+        img1 = gr.Image()
+        text1 = gr.Text()
+    btn1 = gr.Button("button")
+```
+
+以上 gr.Row 定义一行，左边图片，右边输入框。Blocks默认列布局，所以第一列是gr.Row，第二列是一按钮。
+
+
+
+
+```
+with gr.Blocks() as demo:
+    gr.Markdown("Flip text or image files using this demo.")
+    with gr.Tab("Flip Text"):
+        text_input = gr.Textbox()
+        text_output = gr.Textbox()
+        text_button = gr.Button("Flip")
+    with gr.Tab("Flip Image"):
+        with gr.Row():
+            image_input = gr.Image()
+            image_output = gr.Image()
+        image_button = gr.Button("Flip")
+    with gr.Accordion("Open for More!"):
+        gr.Markdown("Look at me...")
+    text_button.click(flip_text, inputs=text_input, outputs=text_output)
+    image_button.click(flip_image, inputs=image_input, outputs=image_output)
+```
+
+以上定义了一个Tab，包含两个选项卡，一个是文本翻转，一个是图片翻转。最后定义按钮点击事件。
+
+
+
+## 样式
+
+img = gr.Image("lion.jpg").style(height='24', rounded=False)
+
+
+## 队列
+
+当请求时间很长，保证websocket连接不断开，开启。
+
+```
+demo = gr.Interface(...).queue()
+# 或
+with gr.Blocks() as demo:
+    pass
+demo.queue()
+```
+
+## 生成器
+
+每隔1s生成一张图片，共生成steps张图片。
+```
+import gradio as gr
+import numpy as np
+import time
+#生成steps张图片，每隔1秒钟返回
+def fake_diffusion(steps):
+    for _ in range(steps):
+        time.sleep(1)
+        image = np.random.randint(255, size=(300, 600, 3))
+        yield image
+demo = gr.Interface(fake_diffusion,
+                    #设置滑窗，最小值为1，最大值为10，初始值为3，每次改动增减1位
+                    inputs=gr.Slider(1, 10, value=3, step=1),
+                    outputs="image")
+#生成器必须要queue函数
+demo.queue()
+demo.launch()
+```
+
+## 会话
+
+在多个web页面中共享状态。
+
+```
+data ={"name": "John", "age": 25}, {"name": "Jane", "age": 30};
+
+
+import gradio as gr
+
+def session_state_demo(input_text, history):
+    # Your processing logic here
+    history.append("Processed: " + input_text)
+    return history
+
+iface_session = gr.Interface(
+    fn=session_state_demo, inputs=["text", gr.Session("text")], outputs="text"
+)
+
+
+```
+
+
+在 inputs 中带入 gr.Session
+
+
+
+
+
 
 
 ## Reference

docs/quick_start.md

@@ -1,6 +1,8 @@
 # 开始开始
 
-安装：
+gradle 快速搭建机器学习web界面，用户输入表单，模型预测，输出结果。所以只需要定义好这些即可。
+
+Python3.8+，安装：
     
 ```bash
 pip install gradio
@@ -22,3 +24,10 @@ interface.launch()
 ```
 
 这里有 Interface 类，接受三个参数：
+
+
+这里 share=True 分享外链，其实是使用frp搭建的内网穿透，使用 gradio 公共服务器进行转发，这里可以使用自己搭建的服务器转发。
+
+
+
+

docs/web.md

@@ -3,5 +3,33 @@
 可集成 Flask、Django和FastAPI
 
 
-##
+## fastapi
+
+首先导入fastapi, gradio，分别新建对象，然后通过 gr.mount_gradio_app 函数，将fastapi挂载到gradio上：
+
+```
+from fastapi import FastAPI
+import gradio as gr
+
+app = FastAPI()
+demo = gr.Interface(
+    fn=gradio_predict,
+    inputs=gr.Textbox(
+        label="Ask a question", placeholder="What is the capital of France?"
+    ),
+    outputs=[gr.Textbox(label="Answer"), gr.Number(label="Score")],
+    allow_flagging="never",
+)
+fastapp = gr.mount_gradio_app(app, demo, path="/")
+```
+
+接着可以分别写 fastapi 和 gradio 各自的功能，比如 fastapi 做权鉴。
+
+## 
+
+
+
+
+
+
 

pyproject.toml

@@ -9,6 +9,8 @@ readme = "README.md"
 python = "^3.12"
 gradio = "^4.32.2"
 opencv-python = "^4.10.0.82"
+transformers = "^4.41.2"
+pygwalker = "^0.4.8.9"
 
 
 [build-system]