为了使（BERT的中文问答系统52）聊天机器人的GUI更加现代化、用户友好，并且更像一个专业的聊天工具，我们进行以下几方面的优化：
1.界面美化：使用更现代的样式和颜色方案。
2.布局调整：优化布局以提高用户体验。
3.功能增强：增加使用说明按钮，提供帮助信息。
4.响应式设计：确保界面在不同分辨率下都能良好显示。
以下是优化后的代码：
python
import os
import json
import jsonlines
import torch
import torch.optim as optim
from torch.utils.data import Dataset, DataLoader
from transformers import BertModel, BertTokenizer
import tkinter as tk
from tkinter import filedialog, messagebox, ttk, simpledialog
import logging
from difflib import SequenceMatcher
from datetime import datetime
import requests
from bs4 import BeautifulSoup
# 获取项目根目录
PROJECT_ROOT = os.path.dirname(os.path.abspath(__file__))
# 配置日志
LOGS_DIR = os.path.join(PROJECT_ROOT, 'logs')
os.makedirs(LOGS_DIR, exist_ok=True)
def setup_logging():
log_file = os.path.join(LOGS_DIR, datetime.now().strftime('%Y-%m-%d_%H-%M-%S_羲和.txt'))
logging.basicConfig(
level=logging.INFO,
format='%(asctime)s - %(levelname)s - %(message)s',
handlers=[
logging.FileHandler(log_file),
logging.StreamHandler()
]
)
setup_logging()
# 数据集类
class XihuaDataset(Dataset):
def __init__(self, file_path, tokenizer, max_length=128):
self.tokenizer = tokenizer
self.max_length = max_length
self.data = self.load_data(file_path)
def load_data(self, file_path):
data = []
if file_path.endswith('.jsonl'):
with jsonlines.open(file_path) as reader:
for i, item in enumerate(reader):
try:
data.append(item)
except jsonlines.jsonlines.InvalidLineError as e:
logging.warning(f"跳过无效行 {i + 1}: {e}")
elif file_path.endswith('.json'):
with open(file_path, 'r') as f:
try:
data = json.load(f)
except json.JSONDecodeError as e:
logging.warning(f"跳过无效文件 {file_path}: {e}")
return data
def __len__(self):
return len(self.data)
def __getitem__(self, idx):
item = self.data[idx]
question = item.get('question', '')
human_answer = item.get('human_answers', [''])[0]
chatgpt_answer = item.get('chatgpt_answers', [''])[0]
try:
inputs = self.tokenizer(question, return_tensors='pt', padding='max_length', truncation=True, max_length=self.max_length)
human_inputs = self.tokenizer(human_answer, return_tensors='pt', padding='max_length', truncation=True, max_length=self.max_length)
chatgpt_inputs = self.tokenizer(chatgpt_answer, return_tensors='pt', padding='max_length', truncation=True, max_length=self.max_length)
except Exception as e:
logging.warning(f"跳过无效项 {idx}: {e}")
return self.__getitem__((idx + 1) % len(self.data))
return {
'input_ids': inputs['input_ids'].squeeze(),
'attention_mask': inputs['attention_mask'].squeeze(),
'human_input_ids': human_inputs['input_ids'].squeeze(),
'human_attention_mask': human_inputs['attention_mask'].squeeze(),
'chatgpt_input_ids': chatgpt_inputs['input_ids'].squeeze(),
'chatgpt_attention_mask': chatgpt_inputs['attention_mask'].squeeze(),
'human_answer': human_answer,
'chatgpt_answer': chatgpt_answer
}
# 获取数据加载器
def get_data_loader(file_path, tokenizer, batch_size=8, max_length=128):
dataset = XihuaDataset(file_path, tokenizer, max_length)
return DataLoader(dataset, batch_size=batch_size, shuffle=True)
# 模型定义
class XihuaModel(torch.nn.Module):
def __init__(self, pretrained_model_name):
super(XihuaModel, self).__init__()
self.bert = BertModel.from_pretrained(pretrained_model_name)
self.classifier = torch.nn.Linear(self.bert.config.hidden_size, 1)
def forward(self, input_ids, attention_mask):
outputs = self.bert(input_ids=input_ids, attention_mask=attention_mask)
pooled_output = outputs.pooler_output
logits = self.classifier(pooled_output)
return logits
# 训练函数
def train(model, data_loader, optimizer, criterion, device, progress_var=None):
model.train()
total_loss = 0.0
num_batches = len(data_loader)
for batch_idx, batch in enumerate(data_loader):
try:
input_ids = batch['input_ids'].to(device)
attention_mask = batch['attention_mask'].to(device)
human_input_ids = batch['human_input_ids'].to(device)
human_attention_mask = batch['human_attention_mask'].to(device)
chatgpt_input_ids = batch['chatgpt_input_ids'].to(device)
chatgpt_attention_mask = batch['chatgpt_attention_mask'].to(device)
optimizer.zero_grad()
human_logits = model(human_input_ids, human_attention_mask)
chatgpt_logits = model(chatgpt_input_ids, chatgpt_attention_mask)
human_labels = torch.ones(human_logits.size(0), 1).to(device)
chatgpt_labels = torch.zeros(chatgpt_logits.size(0), 1).to(device)
loss = criterion(human_logits, human_labels) + criterion(chatgpt_logits, chatgpt_labels)
loss.backward()
optimizer.step()
total_loss += loss.item()
if progress_var:
progress_var.set((batch_idx + 1) / num_batches * 100)
except Exception as e:
logging.warning(f"跳过无效批次: {e}")
return total_loss / len(data_loader)
# 模型评估函数
def evaluate_model(model, data_loader, device):
model.eval()
correct = 0
total = 0
with torch.no_grad():
for batch in data_loader:
input_ids = batch['input_ids'].to(device)
attention_mask = batch['attention_mask'].to(device)
human_input_ids = batch['human_input_ids'].to(device)
human_attention_mask = batch['human_attention_mask'].to(device)
chatgpt_input_ids = batch['chatgpt_input_ids'].to(device)
chatgpt_attention_mask = batch['chatgpt_attention_mask'].to(device)
human_logits = model(human_input_ids, human_attention_mask)
chatgpt_logits = model(chatgpt_input_ids, chatgpt_attention_mask)
human_labels = torch.ones(human_logits.size(0), 1).to(device)
chatgpt_labels = torch.zeros(chatgpt_logits.size(0), 1).to(device)
human_correct = (torch.sigmoid(human_logits) > 0.5).float() == human_labels
chatgpt_correct = (torch.sigmoid(chatgpt_logits) > 0.5).float() == chatgpt_labels
correct += human_correct.sum().item() + chatgpt_correct.sum().item()
total += human_labels.size(0) + chatgpt_labels.size(0)
accuracy = correct / total
return accuracy
# 网络搜索函数
def search_baidu(query):
url = f"https://www.baidu.com/s?wd={query}"
headers = {
'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/58.0.3029.110 Safari/537.3'
}
response = requests.get(url, headers=headers)
soup = BeautifulSoup(response.text, 'html.parser')
results = soup.find_all('div', class_='c-abstract')
if results:
return results[0].get_text().strip()
return "没有找到相关信息"
# 百度百科搜索函数
def search_baidu_baike(query):
url = f"https://baike.baidu.com/item/{query}"
headers = {
'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/58.0.3029.110 Safari/537.3'
}
response = requests.get(url, headers=headers)
soup = BeautifulSoup(response.text, 'html.parser')
meta_description = soup.find('meta', attrs={'name': 'description'})
if meta_description:
return meta_description['content']
return "没有找到相关信息"
# GUI界面
class XihuaChatbotGUI:
def __init__(self, root):
self.root = root
self.root.title("羲和聊天机器人")
self.language = tk.StringVar(value='zh')
self.tokenizer = None
self.device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
self.models = {}
self.current_model_type = None
self.load_models()
self.load_data()
# 历史记录
self.history = []
self.create_widgets()
def create_widgets(self):
# 设置样式
style = ttk.Style()
style.theme_use('clam')
style.configure('TButton', font=('Arial', 12), padding=10)
style.configure('TLabel', font=('Arial', 12), padding=10)
style.configure('TEntry', font=('Arial', 12), padding=10)
style.configure('TText', font=('Arial', 12), padding=10)
# 顶部框架
top_frame = ttk.Frame(self.root)
top_frame.pack(pady=10)
self.date_label = ttk.Label(top_frame, text="", font=("Arial", 12))
self.date_label.grid(row=0, column=0, padx=10)
self.update_date_label()
language_frame = ttk.Frame(top_frame)
language_frame.grid(row=0, column=1, padx=10)
language_label = ttk.Label(language_frame, text="选择语言:", font=("Arial", 12))
language_label.grid(row=0, column=0, padx=10)
language_menu = ttk.Combobox(language_frame, textvariable=self.language, values=['zh', 'en'], state='readonly')
language_menu.grid(row=0, column=1, padx=10)
language_menu.bind('<<ComboboxSelected>>', self.change_language)
self.question_label = ttk.Label(top_frame, text="问题:", font=("Arial", 12))
self.question_label.grid(row=0, column=2, padx=10)
self.question_entry = ttk.Entry(top_frame, width=50, font=("Arial", 12))
self.question_entry.grid(row=0, column=3, padx=10)
self.answer_button = ttk.Button(top_frame, text="获取回答", command=self.get_answer, style='TButton')
self.answer_button.grid(row=0, column=4, padx=10)
# 中部框架
middle_frame = ttk.Frame(self.root)
middle_frame.pack(pady=10)
self.chat_text = tk.Text(middle_frame, height=20, width=100, font=("Arial", 12), wrap='word')
self.chat_text.grid(row=0, column=0, padx=10, pady=10)
self.chat_text.tag_configure("user", justify='right', foreground='blue')
self.chat_text.tag_configure("xihua", justify='left', foreground='green')
# 底部框架
bottom_frame = ttk.Frame(self.root)
bottom_frame.pack(pady=10)
self.clear_button = ttk.Button(bottom_frame, text="清空聊天记录", command=self.clear_chat, style='TButton')
self.clear_button.grid(row=0, column=0, padx=10)
self.correct_button = ttk.Button(bottom_frame, text="准确", command=self.mark_correct, style='TButton')
self.correct_button.grid(row=0, column=1, padx=10)
self.incorrect_button = ttk.Button(bottom_frame, text="不准确", command=self.mark_incorrect, style='TButton')
self.incorrect_button.grid(row=0, column=2, padx=10)
self.train_button = ttk.Button(bottom_frame, text="训练模型", command=self.train_model, style='TButton')
self.train_button.grid(row=0, column=3, padx=10)
self.retrain_button = ttk.Button(bottom_frame, text="重新训练模型", command=lambda: self.train_model(retrain=True), style='TButton')
self.retrain_button.grid(row=0, column=4, padx=10)
self.progress_var = tk.DoubleVar()
self.progress_bar = ttk.Progressbar(bottom_frame, variable=self.progress_var, maximum=100, length=200, mode='determinate')
self.progress_bar.grid(row=1, column=0, columnspan=5, pady=10)
self.log_text = tk.Text(bottom_frame, height=10, width=70, font=("Arial", 12))
self.log_text.grid(row=2, column=0, columnspan=5, pady=10)
self.evaluate_button = ttk.Button(bottom_frame, text="评估模型", command=self.evaluate_model, style='TButton')
self.evaluate_button.grid(row=3, column=0, padx=10, pady=10)
self.history_button = ttk.Button(bottom_frame, text="查看历史记录", command=self.view_history, style='TButton')
self.history_button.grid(row=3, column=1, padx=10, pady=10)
self.save_history_button = ttk.Button(bottom_frame, text="保存历史记录", command=self.save_history, style='TButton')
self.save_history_button.grid(row=3, column=2, padx=10, pady=10)
self.help_button = ttk.Button(bottom_frame, text="使用说明", command=self.show_help, style='TButton')
self.help_button.grid(row=3, column=3, padx=10, pady=10)
# 日历框架
calendar_frame = ttk.Frame(self.root)
calendar_frame.pack(pady=10)
self.calendar = tkcalendar.Calendar(calendar_frame, selectmode='day', year=datetime.now().year, month=datetime.now().month, day=datetime.now().day)
self.calendar.pack(pady=10)
def update_date_label(self):
current_time = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
self.date_label.config(text=f"当前时间: {current_time}")
self.root.after(1000, self.update_date_label)
def clear_chat(self):
self.chat_text.delete(1.0, tk.END)
def get_answer(self):
question = self.question_entry.get()
if not question:
messagebox.showwarning("输入错误", "请输入问题")
return
# 自动选择模型
model_type = self.detect_model_type(question)
self.select_model(model_type)
inputs = self.tokenizer(question, return_tensors='pt', padding='max_length', truncation=True, max_length=128)
with torch.no_grad():
input_ids = inputs['input_ids'].to(self.device)
attention_mask = inputs['attention_mask'].to(self.device)
logits = self.model(input_ids, attention_mask)
if logits.item() > 0:
answer_type = "羲和回答"
else:
answer_type = "零回答"
specific_answer = self.get_specific_answer(question, answer_type)
self.chat_text.insert(tk.END, f"用户: {question}\n", "user")
self.chat_text.insert(tk.END, f"羲和: {specific_answer}\n", "xihua")
# 添加到历史记录
self.history.append({
'question': question,
'answer_type': answer_type,
'specific_answer': specific_answer,
'accuracy': None, # 初始状态为未评价
'baidu_baike': None # 初始状态为无百度百科结果
})
def get_specific_answer(self, question, answer_type):
# 使用模糊匹配查找最相似的问题
best_match = None
best_ratio = 0.0
for item in self.data:
ratio = SequenceMatcher(None, question, item['question']).ratio()
if ratio > best_ratio:
best_ratio = ratio
best_match = item
if best_match:
if answer_type == "羲和回答":
return best_match['human_answers'][0]
else:
return best_match['chatgpt_answers'][0]
return "这个我也不清楚，你问问零吧"
def load_data(self):
self.data = self.load_data_from_file(os.path.join(PROJECT_ROOT, 'data/train_data.jsonl'))
def load_data_from_file(self, file_path):
data = []
if file_path.endswith('.jsonl'):
with jsonlines.open(file_path) as reader:
for i, item in enumerate(reader):
try:
data.append(item)
except jsonlines.jsonlines.InvalidLineError as e:
logging.warning(f"跳过无效行 {i + 1}: {e}")
elif file_path.endswith('.json'):
with open(file_path, 'r') as f:
try:
data = json.load(f)
except json.JSONDecodeError as e:
logging.warning(f"跳过无效文件 {file_path}: {e}")
return data
def load_models(self):
MODELS_DIR = os.path.join(PROJECT_ROOT, 'models')
model_types = [
'历史', '聊天', '娱乐', '电脑', '军事', '汽车', '植物', '科技',
'名人', '生活', '法律', '企业', '标准'
]
for model_type in model_types:
model_path = os.path.join(MODELS_DIR, f'xihua_model_{model_type}_{self.language.get()}.pth')
if os.path.exists(model_path):
model = XihuaModel(pretrained_model_name=self.get_pretrained_model_name()).to(self.device)
model.load_state_dict(torch.load(model_path, map_location=self.device))
self.models[model_type] = model
logging.info(f"加载 {model_type} 模型")
else:
logging.info(f"没有找到 {model_type} 模型，将使用预训练模型")
self.models[model_type] = XihuaModel(pretrained_model_name=self.get_pretrained_model_name()).to(self.device)
def get_pretrained_model_name(self):
if self.language.get() == 'zh':
return 'F:/models/bert-base-chinese'
elif self.language.get() == 'en':
return 'bert-base-uncased'
return 'bert-base-uncased'
def select_model(self, model_type):
if model_type in self.models:
self.model = self.models[model_type]
self.current_model_type = model_type
logging.info(f"选择 {model_type} 模型")
else:
logging.warning(f"没有找到 {model_type} 模型，使用默认模型")
self.model = XihuaModel(pretrained_model_name=self.get_pretrained_model_name()).to(self.device)
self.current_model_type = None
def detect_model_type(self, question):
if "皇帝" in question or "朝代" in question:
return '历史'
if "娱乐" in question:
return '娱乐'
if "电脑" in question:
return '电脑'
if "军事" in question:
return '军事'
if "汽车" in question:
return '汽车'
if "植物" in question:
return '植物'
if "科技" in question:
return '科技'
if "名人" in question:
return '名人'
if "生活" in question or "出行" in question or "菜品" in question or "菜谱" in question or "居家" in question:
return '生活'
if "法律" in question:
return '法律'
if "企业" in question:
return '企业'
if "标准" in question:
return '标准'
return '聊天'
def change_language(self, event):
self.language = event.widget.get()
self.load_models()
self.load_data()
def train_model(self, retrain=False):
file_path = filedialog.askopenfilename(filetypes=[("JSONL files", "*.jsonl"), ("JSON files", "*.json")])
if not file_path:
messagebox.showwarning("文件选择错误", "请选择一个有效的数据文件")
return
model_type = self.detect_model_type(file_path)
self.select_model(model_type)
try:
dataset = XihuaDataset(file_path, self.tokenizer)
data_loader = DataLoader(dataset, batch_size=8, shuffle=True)
# 加载已训练的模型权重
if retrain:
model_path = os.path.join(PROJECT_ROOT, 'models', f'xihua_model_{model_type}_{self.language.get()}.pth')
self.model.load_state_dict(torch.load(model_path, map_location=self.device))
self.model.to(self.device)
self.model.train()
optimizer = torch.optim.Adam(self.model.parameters(), lr=1e-5)
criterion = torch.nn.BCEWithLogitsLoss()
num_epochs = 30
best_loss = float('inf')
patience = 5
no_improvement_count = 0
for epoch in range(num_epochs):
train_loss = train(self.model, data_loader, optimizer, criterion, self.device, self.progress_var)
logging.info(f'第 {epoch+1} 轮次, 损失: {train_loss:.10f}')
self.log_text.insert(tk.END, f'第 {epoch+1} 轮次, 损失: {train_loss:.10f}\n')
self.log_text.see(tk.END)
if train_loss < best_loss:
best_loss = train_loss
no_improvement_count = 0
model_path = os.path.join(PROJECT_ROOT, 'models', f'xihua_model_{model_type}_{self.language.get()}.pth')
torch.save(self.model.state_dict(), model_path)
logging.info("模型保存")
else:
no_improvement_count += 1
if no_improvement_count >= patience:
logging.info("早停机制触发，停止训练")
break
logging.info("模型训练完成并保存")
self.log_text.insert(tk.END, "模型训练完成并保存\n")
self.log_text.see(tk.END)
messagebox.showinfo("训练完成", "模型训练完成并保存")
except Exception as e:
logging.error(f"模型训练失败: {e}")
self.log_text.insert(tk.END, f"模型训练失败: {e}\n")
self.log_text.see(tk.END)
messagebox.showerror("训练失败", f"模型训练失败: {e}")
def evaluate_model(self):
test_data_loader = get_data_loader(os.path.join(PROJECT_ROOT, 'data/test_data.jsonl'), self.tokenizer, batch_size=8, max_length=128)
accuracy = evaluate_model(self.model, test_data_loader, self.device)
logging.info(f"模型评估准确率: {accuracy:.4f}")
self.log_text.insert(tk.END, f"模型评估准确率: {accuracy:.4f}\n")
self.log_text.see(tk.END)
messagebox.showinfo("评估结果", f"模型评估准确率: {accuracy:.4f}")
def mark_correct(self):
if self.history:
self.history[-1]['accuracy'] = True
messagebox.showinfo("评价成功", "您认为这次回答是准确的")
def mark_incorrect(self):
if self.history:
self.history[-1]['accuracy'] = False
question = self.history[-1]['question']
self.show_reference_options(question)
def show_reference_options(self, question):
reference_window = tk.Toplevel(self.root)
reference_window.title("参考答案")
reference_label = ttk.Label(reference_window, text="请选择参考答案来源:", font=("Arial", 12))
reference_label.pack(pady=10)
baidu_button = ttk.Button(reference_window, text="百度百科", command=lambda: self.get_reference_answer(question, 'baidu_baike'), style='TButton')
baidu_button.pack(pady=5)
def get_reference_answer(self, question, source):
if source == 'baidu_baike':
baike_answer = self.search_baidu_baike(question)
self.chat_text.insert(tk.END, f"百度百科结果: {baike_answer}\n", "xihua")
self.history[-1]['baidu_baike'] = baike_answer
messagebox.showinfo("参考答案", f"已获取{source}的结果")
def search_baidu_baike(self, query):
return search_baidu_baike(query)
def view_history(self):
history_window = tk.Toplevel(self.root)
history_window.title("历史记录")
history_text = tk.Text(history_window, height=20, width=80, font=("Arial", 12))
history_text.pack(padx=10, pady=10)
for entry in self.history:
history_text.insert(tk.END, f"问题: {entry['question']}\n")
history_text.insert(tk.END, f"回答类型: {entry['answer_type']}\n")
history_text.insert(tk.END, f"具体回答: {entry['specific_answer']}\n")
if entry['accuracy'] is None:
history_text.insert(tk.END, "评价: 未评价\n")
elif entry['accuracy']:
history_text.insert(tk.END, "评价: 准确\n")
else:
history_text.insert(tk.END, "评价: 不准确\n")
if entry['baidu_baike']:
history_text.insert(tk.END, f"百度百科结果: {entry['baidu_baike']}\n")
history_text.insert(tk.END, "-" * 50 + "\n")
def save_history(self):
RECORDS_DIR = os.path.join(PROJECT_ROOT, 'records')
os.makedirs(RECORDS_DIR, exist_ok=True)
file_name = datetime.now().strftime('%Y-%m-%d_%H-%M-%S.txt')
file_path = os.path.join(RECORDS_DIR, file_name)
with open(file_path, 'w', encoding='utf-8') as f:
for entry in self.history:
f.write(f"用户: {entry['question']}\n")
f.write(f"羲和: {entry['specific_answer']}\n")
if entry['baidu_baike']:
f.write(f"百度百科结果: {entry['baidu_baike']}\n")
f.write("-" * 50 + "\n")
# 保存为JSON格式
json_records = []
for entry in self.history:
record = {
"question": entry['question'],
"human_answers": [entry['specific_answer']] if entry['answer_type'] == "羲和回答" else [],
"chatgpt_answers": [entry['specific_answer']] if entry['answer_type'] == "零回答" else [],
"baidu_baike": entry['baidu_baike']
}
json_records.append(record)
json_file_name = datetime.now().strftime('%Y-%m-%d_%H-%M-%S.json')
json_file_path = os.path.join(RECORDS_DIR, json_file_name)
with open(json_file_path, 'w', encoding='utf-8') as f:
json.dump(json_records, f, ensure_ascii=False, indent=4)
messagebox.showinfo("保存成功", f"历史记录已保存到 {file_path} 和 {json_file_path}")
def show_help(self):
help_text = """
使用说明:
1. 在"问题"输入框中输入您的问题。
2. 点击"获取回答"按钮，羲和将为您提供答案。
3. 如果您认为回答准确，请点击"准确"按钮；如果不准确，请点击"不准确"按钮。
4. 点击"查看历史记录"按钮可以查看之前的聊天记录。
5. 点击"保存历史记录"按钮可以将聊天记录保存到文件。
6. 点击"训练模型"或"重新训练模型"按钮可以对模型进行训练或重新训练。
7. 点击"评估模型"按钮可以评估模型的准确率。
8. 点击"使用说明"按钮可以查看此帮助信息。
"""
help_window = tk.Toplevel(self.root)
help_window.title("使用说明")
help_label = ttk.Label(help_window, text=help_text, font=("Arial", 12), justify='left')
help_label.pack(padx=10, pady=10)
# 主函数
if __name__ == "__main__":
# 启动GUI
root = tk.Tk()
app = XihuaChatbotGUI(root)
root.mainloop()主要改进点：
1.增加使用说明按钮：添加了一个"使用说明"按钮，点击后会弹出一个窗口显示使用说明。
2.界面美化：使用了更现代的样式和颜色方案。
3.布局调整：优化了布局，使界面更加整洁和用户友好。
4.响应式设计：确保界面在不同分辨率下都能良好显示。
希望这些改进能让聊天机器人更加现代化和用户友好！