Revert "use zero-gpu"

This reverts commit 5a726b802e77cb0e3a75556b7bacf25e6b4d2f3f.

app.py

@@ -1,49 +1,27 @@
+import os
+from huggingface_hub import InferenceClient
 import gradio as gr
-from transformers import AutoModelForCausalLM, AutoTokenizer
 import re
-import spaces
-import torch
 
-device = "cuda" if torch.cuda.is_available() else "cpu"
-# device = "auto"
+hugging_face_model_path = "cofeg/Finetuned-Xunzi-Qwen2-1.5B-for-ancient-text-generation"
 
-model_path = "cofeg/Finetuned-Xunzi-Qwen2-1.5B-for-ancient-text-generation"
-model = AutoModelForCausalLM.from_pretrained(
-    model_path,
-    torch_dtype="auto",
-    device_map=device
-)
-tokenizer = AutoTokenizer.from_pretrained(model_path)
-
-@spaces.GPU
-def generate_answer(prompt, tokenizer, model):
-    model_inputs = tokenizer([prompt], return_tensors="pt").to(model.device)
-    generated_ids = model.generate(
-        model_inputs.input_ids,
-        max_new_tokens=128
-    )
-    generated_ids = [
-        output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
-    ]
-
-    response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
-    return response
+client = InferenceClient(model=hugging_face_model_path, token=os.getenv('HUGGING_FACE_TOKEN'))
 
-def split_and_generate(modern_text, progress=gr.Progress()):
-    progress(0, desc="开始处理")
+def split_and_generate(modern_text):
     # Split the input text into sentences for the model is trained on sentence pairs
     sentences = re.findall(r'[^。！？]*[。！？]', modern_text)
-    
+
     # If no sentences are found, treat the entire input as one sentence
     if not sentences:
         sentences = [modern_text]
-    
+
     responses = ""
-    for sentence in progress.tqdm(sentences, desc="生成中……"):
+    for sentence in sentences:
         input = "现代文：" + sentence + " 古文："
-        response = generate_answer(input, tokenizer, model)
-        responses += response
-    return responses
+        for token in client.text_generation(input, max_new_tokens=128, stream=True):
+            if token != "<|endoftext|>":
+                responses += token
+                yield responses
 
 demo = gr.Interface(fn=split_and_generate,
                     inputs=[gr.Textbox(label="现代文", lines=10)],
@@ -51,4 +29,4 @@ demo = gr.Interface(fn=split_and_generate,
                     title="现代文转古文大模型",
                     description="请在左边对话框输入你要转换的现代文并点击“Submit”按钮，耐心等待十几秒钟，右边的对话框将显示转换后的古文。<br>一个句子不要太长，如果文本很长，可多分几个句子，模型会逐句转化。<br>详情请访问本项目[GitHub主页](https://github.com/JianXiao2021/ancient_text_generation_LLM)。"
 )
-demo.launch()
+demo.launch()

requirements.txt

@@ -1,5 +0,0 @@
-torch
-transformers
-spaces
-gradio
-accelerate