Update app.py

app.py

@@ -2,41 +2,28 @@ import gradio as gr
 from transformers import AutoModel, AutoTokenizer
 import torch
 
-# Load model (cached after first run)
-def load_model():
-    tokenizer = AutoTokenizer.from_pretrained(
-        "Qwen/Qwen3-Embedding-8B",
-        trust_remote_code=True
-    )
-    model = AutoModel.from_pretrained(
-        "Qwen/Qwen3-Embedding-8B",
-        trust_remote_code=True,
-        device_map="auto"
-    ).eval()
-    return tokenizer, model
-
-tokenizer, model = load_model()
+# Load model with caching
+tokenizer = AutoTokenizer.from_pretrained(
+    "Qwen/Qwen3-Embedding-8B",
+    trust_remote_code=True
+)
+model = AutoModel.from_pretrained(
+    "Qwen/Qwen3-Embedding-8B",
+    trust_remote_code=True,
+    device_map="auto"
+).eval()
 
-# Embedding generation function
 def get_embedding(text):
     inputs = tokenizer(text, return_tensors="pt", truncation=True).to(model.device)
     with torch.no_grad():
         outputs = model(**inputs)
-    # Mean pooling for sentence embedding
     embedding = outputs.last_hidden_state.mean(dim=1).squeeze().tolist()
-    return {
-        "input_text": text,
-        "embedding_size": len(embedding),
-        "first_5_values": embedding[:5]  # Preview
-    }
+    return {"text": text, "embedding_size": len(embedding)}
 
-# Gradio Interface
 demo = gr.Interface(
     fn=get_embedding,
-    inputs=gr.Textbox(label="Input Text", placeholder="Enter text to embed..."),
-    outputs=gr.JSON(label="Embedding Result"),
-    title="Qwen3-Embedding-8B Demo",
-    examples=["Hello world", "How does AI work?", "上海天气怎么样？"]
+    inputs=gr.Textbox(label="Input text"),
+    outputs=gr.JSON(),
+    title="Qwen3 Embeddings"
 )
-
 demo.launch()