Update app.py

app.py

@@ -1,22 +1,31 @@
-import torch
+import gradio as gr
 from transformers import AutoModelForCausalLM, AutoTokenizer
-from peft import PeftModel
-
-# 加载基础模型
-base_model = AutoModelForCausalLM.from_pretrained("unsloth/Phi-4-unsloth-bnb-4bit")
-tokenizer = AutoTokenizer.from_pretrained("unsloth/Phi-4-unsloth-bnb-4bit")
 
-# 加载 LoRA 模型
-model = PeftModel.from_pretrained(base_model, "muchuan-l/phi-4-242v2")
+# 加载模型和分词器
+model_name = "muchuan-l/qwen2.5math1.5b-2v2-rev"
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(model_name, device_map="auto")
 
-# 对话生成函数
-def generate_response(prompt):
-    inputs = tokenizer(prompt, return_tensors="pt")
-    outputs = model.generate(**inputs, max_length=100)
-    return tokenizer.decode(outputs[0], skip_special_tokens=True)
+# 定义对话函数
+def chat(input_text):
+    # 将输入文本转换为模型输入
+    inputs = tokenizer(input_text, return_tensors="pt").to("cuda")
+    
+    # 生成回复
+    outputs = model.generate(**inputs, max_new_tokens=100)
+    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
+    
+    return response
 
-# Gradio 界面
-import gradio as gr
+# 创建 Gradio 界面
+interface = gr.Interface(
+    fn=chat,  # 对话函数
+    inputs="text",  # 输入类型
+    outputs="text",  # 输出类型
+    title="Qwen2.5Math1.5B Chat",  # 界面标题
+    description="A chatbot powered by Qwen2.5Math1.5B model.",  # 界面描述
+    examples=["What is 2 + 2?", "Explain the Pythagorean theorem."]  # 示例输入
+)
 
-iface = gr.Interface(fn=generate_response, inputs="text", outputs="text")
-iface.launch()
+# 启动界面
+interface.launch()