Update app.py

app.py

@@ -1,48 +1,50 @@
 import gradio as gr
 from transformers import AutoTokenizer, AutoModelForCausalLM
+from peft import PeftModel, PeftConfig
 import torch
 
-# Load the Pidgin LLaMA model
-MODEL_NAME = "Ephraimmm/pdgn_llama_model"
+# Model paths
+ADAPTER_MODEL = "Ephraimmm/pdgn_llama_model"
 
-print("Loading model and tokenizer...")
-tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
-model = AutoModelForCausalLM.from_pretrained(
-    MODEL_NAME,
-    torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
+print("Loading LoRA adapter configuration...")
+peft_config = PeftConfig.from_pretrained(ADAPTER_MODEL)
+BASE_MODEL = peft_config.base_model_name_or_path
+
+print(f"Base model: {BASE_MODEL}")
+print(f"Adapter model: {ADAPTER_MODEL}")
+
+print("\nLoading tokenizer...")
+tokenizer = AutoTokenizer.from_pretrained(BASE_MODEL, trust_remote_code=True)
+
+if tokenizer.pad_token is None:
+    tokenizer.pad_token = tokenizer.eos_token
+
+print("Loading base model...")
+base_model = AutoModelForCausalLM.from_pretrained(
+    BASE_MODEL,
+    dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
     device_map="auto" if torch.cuda.is_available() else None,
-    low_cpu_mem_usage=True
+    low_cpu_mem_usage=True,
+    trust_remote_code=True
 )
 
+print("Loading LoRA adapter...")
+model = PeftModel.from_pretrained(base_model, ADAPTER_MODEL)
+model.eval()
+
 print("Model loaded successfully!")
 
 def chat_with_pidgin_bot(message, history, system_prompt, max_length=512, temperature=0.7, top_p=0.9):
-    """
-    Generate a response from the Pidgin LLaMA model
-    
-    Args:
-        message: User's input message
-        history: Chat history (list of [user_msg, bot_msg] pairs)
-        system_prompt: System instructions for the chatbot behavior
-        max_length: Maximum length of generated response
-        temperature: Sampling temperature (higher = more random)
-        top_p: Nucleus sampling parameter
-    """
-    
-    # Build conversation context with system prompt
     conversation = f"System: {system_prompt}\n\n" if system_prompt else ""
     for user_msg, bot_msg in history:
         conversation += f"User: {user_msg}\nAssistant: {bot_msg}\n"
     conversation += f"User: {message}\nAssistant:"
     
-    # Tokenize input
     inputs = tokenizer(conversation, return_tensors="pt", truncation=True, max_length=2048)
     
-    # Move to GPU if available
     if torch.cuda.is_available():
         inputs = inputs.to("cuda")
     
-    # Generate response
     with torch.no_grad():
         outputs = model.generate(
             **inputs,
@@ -54,114 +56,56 @@ def chat_with_pidgin_bot(message, history, system_prompt, max_length=512, temper
             eos_token_id=tokenizer.eos_token_id,
         )
     
-    # Decode response
     response = tokenizer.decode(outputs[0], skip_special_tokens=True)
-    
-    # Extract only the assistant's response
     response = response.split("Assistant:")[-1].strip()
     
-    # Remove any potential "User:" that might appear
     if "User:" in response:
         response = response.split("User:")[0].strip()
     
     return response
 
-# Custom CSS for styling
 custom_css = """
 #chatbot {
     height: 500px;
 }
-.gradio-container {
-    font-family: 'Arial', sans-serif;
-}
 """
 
-# Create Gradio interface
 with gr.Blocks(css=custom_css, theme=gr.themes.Soft()) as demo:
     gr.Markdown(
         """
-        # 🗣️ Pidgin LLaMA Chatbot
+        # Pidgin LLaMA Chatbot
         ### Chat with an AI trained on Nigerian Pidgin English
         
-        This chatbot uses the **Ephraimmm/pdgn_llama_model** from Hugging Face.
-        Start a conversation in Pidgin or English!
+        This chatbot uses a LoRA fine-tuned model for Nigerian Pidgin.
         """
     )
     
-    chatbot = gr.Chatbot(
-        label="Pidgin Chat",
-        elem_id="chatbot",
-        bubble_full_width=False,
-        avatar_images=(None, "🤖")
-    )
+    chatbot = gr.Chatbot(label="Pidgin Chat", elem_id="chatbot")
     
     with gr.Row():
-        msg = gr.Textbox(
-            label="Your Message",
-            placeholder="Wetin dey your mind? Type your message here...",
-            scale=4
-        )
-        send_btn = gr.Button("Send 📤", scale=1, variant="primary")
+        msg = gr.Textbox(label="Your Message", placeholder="Type your message here...", scale=4)
+        send_btn = gr.Button("Send", scale=1, variant="primary")
     
-    with gr.Accordion("🎯 System Prompt", open=True):
+    with gr.Accordion("System Prompt", open=True):
         system_prompt = gr.Textbox(
             label="System Instructions",
-            placeholder="Enter system prompt to guide the chatbot's behavior...",
-            value="You are a helpful AI assistant that speaks Nigerian Pidgin English. You are friendly, respectful, and knowledgeable about Nigerian culture. Respond naturally in Pidgin while being helpful and informative.",
-            lines=4,
-            max_lines=8
-        )
-        
-        gr.Markdown(
-            """
-            **Preset Prompts:**
-            """
+            value="You are a helpful AI assistant that speaks Nigerian Pidgin English. You are friendly, respectful, and knowledgeable about Nigerian culture.",
+            lines=4
         )
         
         with gr.Row():
-            preset1 = gr.Button("🎭 Comedian", size="sm")
-            preset2 = gr.Button("📚 Teacher", size="sm")
-            preset3 = gr.Button("🤝 Friend", size="sm")
-            preset4 = gr.Button("💼 Professional", size="sm")
-    
-    with gr.Accordion("⚙️ Advanced Settings", open=False):
-        max_length = gr.Slider(
-            minimum=50,
-            maximum=1024,
-            value=512,
-            step=50,
-            label="Max Response Length",
-            info="Maximum number of tokens to generate"
-        )
-        temperature = gr.Slider(
-            minimum=0.1,
-            maximum=2.0,
-            value=0.7,
-            step=0.1,
-            label="Temperature",
-            info="Higher values make output more random"
-        )
-        top_p = gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.9,
-            step=0.05,
-            label="Top P",
-            info="Nucleus sampling parameter"
-        )
+            preset1 = gr.Button("Comedian")
+            preset2 = gr.Button("Teacher")
+            preset3 = gr.Button("Friend")
+            preset4 = gr.Button("Professional")
     
-    clear = gr.Button("Clear Chat 🗑️")
+    with gr.Accordion("Advanced Settings", open=False):
+        max_length = gr.Slider(50, 1024, 512, step=50, label="Max Response Length")
+        temperature = gr.Slider(0.1, 2.0, 0.7, step=0.1, label="Temperature")
+        top_p = gr.Slider(0.1, 1.0, 0.9, step=0.05, label="Top P")
     
-    gr.Markdown(
-        """
-        ### Example Questions:
-        - "How far? Wetin dey happen?"
-        - "Tell me about Nigerian culture"
-        - "Na wetin be your name?"
-        """
-    )
+    clear = gr.Button("Clear Chat")
     
-    # Event handlers
     def respond(message, chat_history, sys_prompt, max_len, temp, top_p_val):
         bot_message = chat_with_pidgin_bot(message, chat_history, sys_prompt, max_len, temp, top_p_val)
         chat_history.append((message, bot_message))
@@ -169,24 +113,15 @@ with gr.Blocks(css=custom_css, theme=gr.themes.Soft()) as demo:
     
     def set_preset(preset_type):
         presets = {
-            "comedian": "You are a Nigerian comedian who speaks Pidgin. Make people laugh with your witty responses and funny observations about everyday life. Use humor, wordplay, and relatable Nigerian experiences.",
-            "teacher": "You are a patient and knowledgeable teacher who speaks Nigerian Pidgin. Explain things clearly and make learning fun. Use examples from Nigerian culture and everyday life to help students understand.",
-            "friend": "You are a caring and supportive friend who speaks Nigerian Pidgin. Listen to people's problems, give good advice, and always be there for them. Be empathetic, encouraging, and real.",
-            "professional": "You are a professional business consultant who speaks Nigerian Pidgin. Provide clear, practical advice on business and career matters. Be respectful, knowledgeable, and solution-oriented."
+            "comedian": "You are a Nigerian comedian who speaks Pidgin. Make people laugh with witty responses.",
+            "teacher": "You are a patient teacher who speaks Nigerian Pidgin. Explain things clearly.",
+            "friend": "You are a caring friend who speaks Nigerian Pidgin. Give good advice.",
+            "professional": "You are a professional consultant who speaks Nigerian Pidgin. Provide practical advice."
         }
         return presets.get(preset_type, "")
     
-    msg.submit(
-        respond,
-        [msg, chatbot, system_prompt, max_length, temperature, top_p],
-        [msg, chatbot]
-    )
-    
-    send_btn.click(
-        respond,
-        [msg, chatbot, system_prompt, max_length, temperature, top_p],
-        [msg, chatbot]
-    )
+    msg.submit(respond, [msg, chatbot, system_prompt, max_length, temperature, top_p], [msg, chatbot])
+    send_btn.click(respond, [msg, chatbot, system_prompt, max_length, temperature, top_p], [msg, chatbot])
     
     preset1.click(lambda: set_preset("comedian"), None, system_prompt)
     preset2.click(lambda: set_preset("teacher"), None, system_prompt)
@@ -195,10 +130,5 @@ with gr.Blocks(css=custom_css, theme=gr.themes.Soft()) as demo:
     
     clear.click(lambda: None, None, chatbot, queue=False)
 
-# Launch the app
 if __name__ == "__main__":
-    demo.launch(
-        share=True,  # Creates a public link
-        server_name="0.0.0.0",  # Makes it accessible on your network
-        server_port=7860
-    )
+    demo.launch(share=True, server_name="0.0.0.0", server_port=7860)