fixing

app.py

@@ -2,7 +2,7 @@ import os
 import warnings
 import torch
 import gc
-from transformers import AutoModelForVision2Seq, AutoProcessor, BitsAndBytesConfig, AutoConfig
+from transformers import pipeline, AutoTokenizer
 from PIL import Image
 import gradio as gr
 from huggingface_hub import login
@@ -11,65 +11,43 @@ warnings.filterwarnings('ignore')
 os.environ["CUDA_VISIBLE_DEVICES"] = "0"
 
 # Global variables
-model = None
-processor = None
+pipe = None
 
 if torch.cuda.is_available():
     torch.cuda.empty_cache()
     gc.collect()
     print("เคลียร์ CUDA cache เรียบร้อยแล้ว")
 
-def load_model_and_processor():
-    """โหลดโมเดลและ processor"""
-    global model, processor
-    print("กำลังโหลดโมเดลและ processor...")
+def load_pipeline():
+    """โหลด pipeline"""
+    global pipe
+    print("กำลังโหลด pipeline...")
     
     try:
-        base_model_path = "meta-llama/Llama-3.2-11B-Vision-Instruct"
         hub_model_path = "Aekanun/thai-handwriting-llm"
         
-        # Load and set config
-        config = AutoConfig.from_pretrained(
-            hub_model_path,
-            trust_remote_code=True,
+        # สร้าง pipeline
+        pipe = pipeline(
+            "image-to-text",
+            model=hub_model_path,
+            device="cuda" if torch.cuda.is_available() else "cpu",
+            model_kwargs={
+                "torch_dtype": torch.bfloat16,
+                "load_in_4bit": True,
+                "trust_remote_code": True,
+            },
             token=os.environ.get('HUGGING_FACE_HUB_TOKEN')
         )
-        config.model_type = "vision2seq"
-
-        bnb_config = BitsAndBytesConfig(
-            load_in_4bit=True,
-            bnb_4bit_use_double_quant=True,
-            bnb_4bit_quant_type="nf4",
-            bnb_4bit_compute_dtype=torch.bfloat16
-        )
-        
-        # โหลด processor จาก base model
-        print("Loading processor...")
-        processor = AutoProcessor.from_pretrained(
-            base_model_path, 
-            token=os.environ.get('HUGGING_FACE_HUB_TOKEN')
-        )
-        
-        # โหลดโมเดลจาก Hub
-        print("Loading model...")
-        model = AutoModelForVision2Seq.from_pretrained(
-            hub_model_path,
-            config=config,
-            device_map="auto",
-            torch_dtype=torch.bfloat16,
-            quantization_config=bnb_config,
-            token=os.environ.get('HUGGING_FACE_HUB_TOKEN')
-        )
-        print("Model loaded successfully!")
         
+        print("โหลด pipeline สำเร็จ!")
         return True
     except Exception as e:
-        print(f"เกิดข้อผิดพลาดในการโหลดโมเดล: {str(e)}")
+        print(f"เกิดข้อผิดพลาดในการโหลด pipeline: {str(e)}")
         return False
 
 def process_handwriting(image):
     """ฟังก์ชันสำหรับ Gradio interface"""
-    global model, processor
+    global pipe
     
     if image is None:
         return "กรุณาอัพโหลดรูปภาพ"
@@ -83,40 +61,26 @@ def process_handwriting(image):
         if image.mode != "RGB":
             image = image.convert("RGB")
 
-        prompt = """Transcribe the Thai handwritten text from the provided image.
-Only return the transcription in Thai language."""
-
-        messages = [
-            {
-                "role": "user",
-                "content": [
-                    {"type": "text", "text": prompt},
-                    {"type": "image", "image": image}
-                ],
-            }
-        ]
-
-        text = processor.apply_chat_template(messages, tokenize=False)
-        inputs = processor(text=text, images=image, return_tensors="pt")
-        inputs = {k: v.to(model.device) for k, v in inputs.items()}
-
-        with torch.no_grad():
-            outputs = model.generate(
-                **inputs,
-                max_new_tokens=256,
-                do_sample=False,
-                pad_token_id=processor.tokenizer.pad_token_id
-            )
-
-        transcription = processor.decode(outputs[0], skip_special_tokens=True)
-        return transcription.strip()
+        # ใช้ pipeline ประมวลผล
+        result = pipe(
+            image,
+            prompt="""Transcribe the Thai handwritten text from the provided image.
+Only return the transcription in Thai language.""",
+            max_new_tokens=256,
+            do_sample=False
+        )
+        
+        # รับผลลัพธ์
+        if isinstance(result, list):
+            return result[0]['generated_text'].strip()
+        return result['generated_text'].strip()
         
     except Exception as e:
         return f"เกิดข้อผิดพลาด: {str(e)}"
 
 # Initialize application
 print("กำลังเริ่มต้นแอปพลิเคชัน...")
-if load_model_and_processor():
+if load_pipeline():
     # Create Gradio interface
     demo = gr.Interface(
         fn=process_handwriting,