Update app.py

app.py

@@ -1,64 +1,83 @@
 import gradio as gr
-from huggingface_hub import InferenceClient
-
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
-
-
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
-
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
-
-    messages.append({"role": "user", "content": message})
-
-    response = ""
-
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
-
-        response += token
-        yield response
-
-
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
-demo = gr.ChatInterface(
-    respond,
-    additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
-    ],
+import os
+import torch
+from unsloth import FastLanguageModel
+from huggingface_hub import spaces
+
+# Get Hugging Face token from environment variables
+HF_TOKEN = os.environ.get('HF_TOKEN')
+
+# Check if we're running in a Hugging Face Space with GPU constraints
+IS_SPACES_ZERO = os.environ.get("SPACES_ZERO_GPU", "0") == "1"
+IS_SPACE = os.environ.get("SPACE_ID", None) is not None
+
+# Determine device (use GPU if available)
+device = "cuda" if torch.cuda.is_available() else "cpu"
+LOW_MEMORY = os.getenv("LOW_MEMORY", "0") == "1"
+
+print(f"Using device: {device}")
+print(f"Low memory mode: {LOW_MEMORY}")
+
+# Model configuration
+max_seq_length = 2048  # Max sequence length for RoPE scaling
+dtype = torch.float16 if device == "cuda" else torch.float32
+load_in_4bit = True  # Enable 4-bit quantization if memory is limited
+
+# Load model and tokenizer with device mapping
+model_name = "nafisneehal/chandler_bot"
+model, tokenizer = FastLanguageModel.from_pretrained(
+    model_name=model_name,
+    max_seq_length=max_seq_length,
+    dtype=dtype,
+    load_in_4bit=load_in_4bit,
+    device_map="auto" if device == "cuda" else None  # Automatic GPU mapping
 )
+FastLanguageModel.for_inference(model)  # Optimize model for faster inference
+
+# Define prompt structure (update if necessary for your model)
+alpaca_prompt = "{instruction} {input} {output}"
+
+instruction_text = "Learn how to talk like Chandler - a popular character from FRIENDS TV Show. Input is someone saying something, Output is what Chandler saying in response."
+
+
+@spaces.GPU  # Use GPU provided by Hugging Face Spaces if available
+def generate_response(user_input, chat_history):
+    instruction = user_input  # Treats user input as instruction
+    input_text = ""  # Any additional input if needed; empty otherwise
+
+    # Prepare inputs for model inference on the correct device
+    inputs = tokenizer(
+        [alpaca_prompt.format(instruction, input_text, "")],
+        return_tensors="pt"
+    ).to(device)  # Ensure tensors are on the correct device
+
+    # Generate response on GPU or CPU as appropriate
+    with torch.no_grad():
+        outputs = model.generate(**inputs, max_new_tokens=100)
+
+    # Decode response
+    bot_reply = tokenizer.decode(outputs[0], skip_special_tokens=True)
+
+    # Update chat history with user and bot interactions
+    chat_history.append(("User", user_input))
+    chat_history.append(("Bot", bot_reply))
+
+    return chat_history, ""  # Returns updated chat history and clears input
+
+
+# Set up Gradio interface
+with gr.Blocks() as demo:
+    gr.Markdown("# Llama-Based Chatbot on GPU")
+
+    chat_history = gr.Chatbot(label="Chat History")
+    user_input = gr.Textbox(
+        placeholder="Type your message here...", label="Your Message")
 
+    # Connect submit actions to generate response function
+    user_input.submit(generate_response, [user_input, chat_history], [
+                      chat_history, user_input])
+    submit_btn = gr.Button("Send")
+    submit_btn.click(generate_response, [user_input, chat_history], [
+                     chat_history, user_input])
 
-if __name__ == "__main__":
-    demo.launch()
+demo.launch()