锦中融合门户系统

# 导入必要的库

from transformers import pipeline

# 加载预训练的大模型

assistant = pipeline("conversational", model="microsoft/DialoGPT-medium")

# 定义对话函数

def chat_with_assistant():

print("智能体助手已启动，输入 'exit' 结束对话。")

while True:

user_input = input("你: ")

if user_input.lower() == "exit":

break

response = assistant(user_input)

print(f"助手: {response[0]['generated_text']}")

# 启动对话

chat_with_assistant()

from transformers import AutoTokenizer, AutoModelForCausalLM, Trainer, TrainingArguments

from datasets import load_dataset

# 加载预训练模型和分词器

model_name = "microsoft/DialoGPT-medium"

tokenizer = AutoTokenizer.from_pretrained(model_name)

model = AutoModelForCausalLM.from_pretrained(model_name)

# 加载学院相关的对话数据

dataset = load_dataset("json", data_files="college_qa.json")

# 对数据进行预处理

def tokenize_function(examples):

return tokenizer(examples["text"], truncation=True, padding="max_length", max_length=512)

tokenized_datasets = dataset.map(tokenize_function, batched=True)

# 设置训练参数

training_args = TrainingArguments(

output_dir="./results",

num_train_epochs=3,

per_device_train_batch_size=4,

save_steps=10_000,

save_total_limit=2,

)

# 定义训练器

trainer = Trainer(

model=model,

args=training_args,

train_dataset=tokenized_datasets["train"],

)

# 开始训练

trainer.train()