Update app.py

app.py

@@ -1,31 +1,40 @@
 import gradio as gr
-from transformers import BertTokenizerFast
+from transformers import AutoTokenizer
 from optimum.onnxruntime import ORTModelForSeq2SeqLM
-from translation_pipeline import TranslationPipeline
 from translator import Translator
+from translation_pipeline import TranslationPipeline
 
-model_id = "hon9kon9ize/bart-translation-zh-yue-onnx"
-tokenizer = BertTokenizerFast.from_pretrained(model_id)
-model = ORTModelForSeq2SeqLM.from_pretrained(model_id, use_cache=False)
-
-pipe = TranslationPipeline(model=model, tokenizer=tokenizer)
-translator = Translator(pipe, batch_size=2, max_length=1024)
+tokenizer = AutoTokenizer.from_pretrained("hon9kon9ize/bart-translation-zh-yue-onnx")
+model = ORTModelForSeq2SeqLM.from_pretrained(
+    "hon9kon9ize/bart-translation-zh-yue-onnx",
+    provider="CPUExecutionProvider",
+    encoder_file_name="encoder_model_quantized.onnx",
+    decoder_file_name="decoder_model_quantized.onnx",
+    decoder_file_with_past_name="decoder_with_past_model_quantized.onnx",
+)
+pipe = TranslationPipeline(model, tokenizer)
+translator = Translator(pipe, max_length=512, batch_size=1)
 
 
-def translate(zh: str):
-    return translator([zh])[0]
+def demo_process(input_text):
+    return translator(input_text)[0]
 
 
 demo = gr.Interface(
-    fn=translate,
+    fn=demo_process,
     inputs=[
         gr.Textbox(label="官話", type="text"),
     ],
     outputs=[
         gr.Textbox(label="廣東話", type="text"),
     ],
-    examples=[["瞧瞧你说的是人话吗？"], ["余文乐关掉潮店被嘲临走还坑人"]],
+    title=f"Chinese to Cantonese Translator",
+    description="This is a demo of the Chinese to Cantonese Translator.",
+    examples=[
+        [
+            "近年成为许多港人热门移居地的英国中部城巿诺定咸（又译诺丁汉，Nottingham），多年来一直面对财政困境，市议会周三（11月29日）宣布破产，是继英国第二大城市伯明翰今年9月宣布破产后，近期「爆煲」的另一个英国主要城市。诺定咸除了维持法例规定必须提供的服务外，巿政府将暂停所有非必要的公共开支。"
+        ]
+    ],
+    cache_examples=True,
 )
-
-if __name__ == "__main__":
-    demo.launch(show_api=False)
+demo.launch(server_name="0.0.0.0")