Upload app.py with huggingface_hub

app.py

@@ -3,25 +3,23 @@ import gradio as gr
 from transformers import pipeline
 
 # -------------------------------
-# Load TinyLLaMA Model
+# Load Model (CPU-safe)
 # -------------------------------
 pipe = pipeline(
     "text-generation",
     model="TinyLlama/TinyLlama-1.1B-Chat-v1.0",
-    torch_dtype=torch.float32,  # CPU-friendly
-    device_map="auto",
+    torch_dtype=torch.float32,
+    device_map=None,
 )
 
 # -------------------------------
-# Chat Function
+# Chat Function (messages format)
 # -------------------------------
 def chat(user_message, history, system_prompt, temperature, max_tokens):
-    # Stronger personality: always prepend system prompt
     messages = [{"role": "system", "content": system_prompt}]
 
-    for u, a in history:
-        messages.append({"role": "user", "content": u})
-        messages.append({"role": "assistant", "content": a})
+    if history:
+        messages.extend(history)
 
     messages.append({"role": "user", "content": user_message})
 
@@ -33,14 +31,17 @@ def chat(user_message, history, system_prompt, temperature, max_tokens):
 
     output = pipe(
         prompt,
-        max_new_tokens=max_tokens,
+        max_new_tokens=int(max_tokens),
         temperature=temperature,
         top_p=0.9,
         do_sample=True,
     )
 
-    response = output[0]["generated_text"].split("<|assistant|>")[-1].strip()
-    history.append([user_message, response])
+    assistant_reply = output[0]["generated_text"].split("<|assistant|>")[-1].strip()
+
+    history.append({"role": "user", "content": user_message})
+    history.append({"role": "assistant", "content": assistant_reply})
+
     return history
 
 # -------------------------------
@@ -48,58 +49,53 @@ def chat(user_message, history, system_prompt, temperature, max_tokens):
 # -------------------------------
 with gr.Blocks(title="TinyLLaMA Chatbot") as demo:
     gr.Markdown("## 🦙 TinyLLaMA Chatbot")
-    gr.Markdown("Select personality to change chatbot behavior and adjust generation settings.")
 
-    # Predefined personalities
     preset_prompts = {
-        "Pirate 🏴‍☠️": "You are a friendly chatbot who always responds like a pirate. Use pirate words like 'Ahoy', 'Matey', and speak in a fun, adventurous style.",
-        "Teacher 👨‍🏫": "You are a patient teacher who explains concepts clearly. Give detailed, step-by-step explanations, and use examples.",
-        "Coder 👨‍💻": "You are a helpful programming assistant. Provide clear code examples and explain syntax professionally.",
-        "Friendly Assistant 🤖": "You are a cheerful and friendly AI assistant. Always respond in a kind, helpful, and approachable tone."
+        "Pirate 🏴‍☠️": "You are a pirate chatbot. Speak like a pirate.",
+        "Teacher 👨‍🏫": "You are a patient teacher.",
+        "Coder 👨‍💻": "You are a programming assistant.",
+        "Friendly 🤖": "You are a friendly assistant."
     }
 
-    system_prompt_dropdown = gr.Dropdown(
+    personality = gr.Dropdown(
         choices=list(preset_prompts.keys()),
         value="Pirate 🏴‍☠️",
-        label="Choose Personality",
+        label="Choose Personality"
     )
 
-    system_prompt_textbox = gr.Textbox(
+    system_prompt = gr.Textbox(
         value=preset_prompts["Pirate 🏴‍☠️"],
-        label="System Prompt (Editable)"
+        label="System Prompt"
     )
 
-    # Update textbox when dropdown changes
-    def update_prompt(choice):
-        return preset_prompts[choice]
-
-    system_prompt_dropdown.change(update_prompt, inputs=system_prompt_dropdown, outputs=system_prompt_textbox)
+    personality.change(
+        lambda x: preset_prompts[x],
+        inputs=personality,
+        outputs=system_prompt
+    )
 
-    chatbot = gr.Chatbot(height=400)
+    chatbot = gr.Chatbot(type="messages", height=400)
     user_input = gr.Textbox(label="Your Message")
 
-    # Generation sliders
-    temperature_slider = gr.Slider(0.1, 1.2, value=0.85, step=0.05, label="Temperature")
-    max_tokens_slider = gr.Slider(32, 256, value=128, step=16, label="Max Tokens")
+    temperature = gr.Slider(0.1, 1.2, value=0.85)
+    max_tokens = gr.Slider(32, 128, value=96, step=16)
 
-    send_btn = gr.Button("Send 🚀")
-    clear_btn = gr.Button("Clear 🧹")
+    send = gr.Button("Send 🚀")
+    clear = gr.Button("Clear 🧹")
 
-    send_btn.click(
+    send.click(
         chat,
-        inputs=[user_input, chatbot, system_prompt_textbox, temperature_slider, max_tokens_slider],
+        inputs=[user_input, chatbot, system_prompt, temperature, max_tokens],
         outputs=chatbot
     )
 
     user_input.submit(
         chat,
-        inputs=[user_input, chatbot, system_prompt_textbox, temperature_slider, max_tokens_slider],
+        inputs=[user_input, chatbot, system_prompt, temperature, max_tokens],
         outputs=chatbot
     )
 
-    clear_btn.click(lambda: [], outputs=chatbot)
+    clear.click(lambda: [], outputs=chatbot)
+
+demo.launch()
 
-# -------------------------------
-# Launch
-# -------------------------------
-demo.launch(share=True)