Update app.py

app.py

@@ -1,118 +1,110 @@
-import gradio as gr
-import torch
-from transformers import AutoModelForCausalLM, AutoTokenizer
-
-# HuatuoGPT model
-model_name = "FreedomIntelligence/HuatuoGPT-7B"
-
-
-class MedicalAI:
-    def __init__(self):
-        self.model = None
-        self.tokenizer = None
-        self.loaded = False
-
-    def load_model(self):
-        if self.loaded:
-            return True
-
-        try:
-            print("🔄 Loading HuatuoGPT-7B...")
-            self.tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
-            self.model = AutoModelForCausalLM.from_pretrained(
-                model_name,
-                torch_dtype=torch.float16,
-                device_map="auto",
-                trust_remote_code=True
-            )
-            self.loaded = True
-            print("✅ HuatuoGPT Ready!")
-            return True
-        except Exception as e:
-            print(f"❌ Error: {e}")
-            return False
-
-
-medical_ai = MedicalAI()
-
-
-def get_ai_thinking(message):
-    """Get HuatuoGPT's complete reasoning process"""
-    if not medical_ai.loaded:
-        success = medical_ai.load_model()
-        if not success:
-            return "🔄 Loading AI... Please wait."
-
-    try:
-        # Simple one-line prompt to see natural reasoning
-        prompt = f"Question: {message}\nAnswer:"
-
-        inputs = medical_ai.tokenizer(prompt, return_tensors="pt")
-
-        # Generate longer response to see full reasoning
-        with torch.no_grad():
-            outputs = medical_ai.model.generate(
-                inputs.input_ids,
-                max_new_tokens=400,  # More tokens to see reasoning
-                temperature=0.7,
-                do_sample=True,
-                top_p=0.9,
-                repetition_penalty=1.1,
-                pad_token_id=medical_ai.tokenizer.eos_token_id
-            )
-
-        # Get full response including reasoning
-        full_response = medical_ai.tokenizer.decode(outputs[0], skip_special_tokens=True)
-
-        # Extract everything after "Answer:"
-        if "Answer:" in full_response:
-            reasoning = full_response.split("Answer:")[1].strip()
-        else:
-            reasoning = full_response.replace(prompt, "").strip()
-
-        return reasoning
-
-    except Exception as e:
-        return f"❌ Error: {str(e)}"
-
-
-# Clean interface focused on reasoning
-with gr.Blocks(theme=gr.themes.Soft()) as demo:
-    gr.Markdown("# 🧠 HuatuoGPT Medical Reasoning")
-    gr.Markdown("**See how the AI doctor thinks and reasons**")
-
-    with gr.Row():
-        with gr.Column(scale=1):
-            input_box = gr.Textbox(
-                label="Your Medical Question",
-                placeholder="headache? fever? sleep problems?",
-                max_lines=1,  # Single line only
-                show_copy_button=True
-            )
-            ask_btn = gr.Button("🤔 Ask HuatuoGPT", variant="primary")
-
-        with gr.Column(scale=2):
-            output_box = gr.Textbox(
-                label="HuatuoGPT's Reasoning",
-                placeholder="AI thinking will appear here...",
-                lines=8,
-                show_copy_button=True
-            )
-
-    clear_btn = gr.Button("🗑️ Clear")
-
-
-    # Function to get AI reasoning
-    def show_reasoning(message):
-        if not message.strip():
-            return "Please enter a medical question."
-        return get_ai_thinking(message)
-
-
-    # Connect actions
-    input_box.submit(show_reasoning, inputs=input_box, outputs=output_box)
-    ask_btn.click(show_reasoning, inputs=input_box, outputs=output_box)
-    clear_btn.click(lambda: ("", ""), outputs=[input_box, output_box])
-
-if __name__ == "__main__":
+import gradio as gr
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+# HuatuoGPT model
+model_name = "FreedomIntelligence/HuatuoGPT-7B"
+
+class MedicalAI:
+    def __init__(self):
+        self.model = None
+        self.tokenizer = None
+        self.loaded = False
+    
+    def load_model(self):
+        if self.loaded:
+            return True
+        
+        try:
+            print("🔄 Loading HuatuoGPT-7B...")
+            self.tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
+            self.model = AutoModelForCausalLM.from_pretrained(
+                model_name,
+                torch_dtype=torch.float16,
+                device_map="auto",
+                trust_remote_code=True
+            )
+            self.loaded = True
+            print("✅ HuatuoGPT Ready!")
+            return True
+        except Exception as e:
+            print(f"❌ Error: {e}")
+            return False
+
+medical_ai = MedicalAI()
+
+def get_ai_thinking(message):
+    """Get HuatuoGPT's complete reasoning process"""
+    if not medical_ai.loaded:
+        success = medical_ai.load_model()
+        if not success:
+            return "🔄 Loading AI... Please wait."
+    
+    try:
+        # Simple one-line prompt to see natural reasoning
+        prompt = f"Question: {message}\nAnswer:"
+        
+        inputs = medical_ai.tokenizer(prompt, return_tensors="pt")
+        
+        # Generate longer response to see full reasoning
+        with torch.no_grad():
+            outputs = medical_ai.model.generate(
+                inputs.input_ids,
+                max_new_tokens=400,  # More tokens to see reasoning
+                temperature=0.7,
+                do_sample=True,
+                top_p=0.9,
+                repetition_penalty=1.1,
+                pad_token_id=medical_ai.tokenizer.eos_token_id
+            )
+        
+        # Get full response including reasoning
+        full_response = medical_ai.tokenizer.decode(outputs[0], skip_special_tokens=True)
+        
+        # Extract everything after "Answer:"
+        if "Answer:" in full_response:
+            reasoning = full_response.split("Answer:")[1].strip()
+        else:
+            reasoning = full_response.replace(prompt, "").strip()
+        
+        return reasoning
+        
+    except Exception as e:
+        return f"❌ Error: {str(e)}"
+
+# FIXED: Remove theme parameter
+with gr.Blocks() as demo:  # Removed: theme=gr.themes.Soft()
+    gr.Markdown("# 🧠 HuatuoGPT Medical Reasoning")
+    gr.Markdown("**See how the AI doctor thinks and reasons**")
+    
+    with gr.Row():
+        with gr.Column(scale=1):
+            input_box = gr.Textbox(
+                label="Your Medical Question",
+                placeholder="headache? fever? sleep problems?",
+                max_lines=1,  # Single line only
+            )
+            ask_btn = gr.Button("🤔 Ask HuatuoGPT", variant="primary")
+        
+        with gr.Column(scale=2):
+            output_box = gr.Textbox(
+                label="HuatuoGPT's Reasoning",
+                placeholder="AI thinking will appear here...",
+                lines=8,
+            )
+    
+    clear_btn = gr.Button("🗑️ Clear")
+    
+    # Function to get AI reasoning
+    def show_reasoning(message):
+        if not message.strip():
+            return "Please enter a medical question."
+        return get_ai_thinking(message)
+    
+    # Connect actions
+    input_box.submit(show_reasoning, inputs=input_box, outputs=output_box)
+    ask_btn.click(show_reasoning, inputs=input_box, outputs=output_box)
+    clear_btn.click(lambda: ("", ""), outputs=[input_box, output_box])
+
+if __name__ == "__main__":
     demo.launch()