Update app.py

app.py

@@ -1,27 +1,24 @@
 import os
 import gradio as gr
+import spaces
 from transformers import AutoTokenizer, AutoModelForCausalLM, TextGenerationPipeline
 
-# 使用你在 Spaces 中添加的 Secret
+model_id = "google/gemma-3-12b-it"
 hf_token = os.environ.get("HUGGINGFACE_TOKEN")
 
-# 加载 tokenizer 和模型（带token）
-model_id = "google/gemma-3-4b-it"
-tokenizer = AutoTokenizer.from_pretrained(model_id, token=hf_token)
-model = AutoModelForCausalLM.from_pretrained(model_id, token=hf_token, device_map="auto")
-
-# 构造推理管道
-pipe = TextGenerationPipeline(model=model, tokenizer=tokenizer)
-
-# 推理函数
+# 包含模型加载 + 推理
+@spaces.GPU
 def generate(prompt):
-    output = pipe(prompt, max_new_tokens=100, do_sample=True, temperature=0.7)
+    tokenizer = AutoTokenizer.from_pretrained(model_id, token=hf_token)
+    model = AutoModelForCausalLM.from_pretrained(model_id, token=hf_token, device_map="auto")
+    pipeline = TextGenerationPipeline(model=model, tokenizer=tokenizer)
+    output = pipeline(prompt, max_new_tokens=100, do_sample=True, temperature=0.7)
     return output[0]["generated_text"]
 
-# Gradio UI
+# 构建界面
 gr.Interface(
     fn=generate,
     inputs=gr.Text(label="Enter your prompt"),
     outputs=gr.Textbox(label="Generated Text"),
-    title="Gemma-3-27B Text Generation"
+    title="Gemma-3-27B Inference (ZeroGPU)"
 ).launch()