Update

app.py

@@ -1,61 +1,64 @@
-import torch
-from transformers import AutoModelForCausalLM, AutoTokenizer
-from peft import PeftModel
 import gradio as gr
+from huggingface_hub import InferenceClient
 
-# Load the base model
-base_model_name = "unsloth/llama-3.2-3b-instruct-bnb-4bit"
-tokenizer = AutoTokenizer.from_pretrained(base_model_name, use_fast=False)
-base_model = AutoModelForCausalLM.from_pretrained(
-    base_model_name,
-    device_map="auto",  # Automatically map layers to available devices
-    torch_dtype=torch.float16  # Ensure compatibility with 4-bit quantization
-)
+"""
+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("Grandediw/lora_model")
 
-# Load the LoRA adapter
-adapter_path = "Grandediw/lora_model"  # Replace with your model path
-model = PeftModel.from_pretrained(base_model, adapter_path)
-model.eval()  # Set the model to evaluation mode
 
-# Define the inference function
 def respond(
     message,
     history: list[tuple[str, str]],
+    system_message,
     max_tokens,
     temperature,
     top_p,
 ):
-    # Build context from history
-    context = ""
-    for user_message, assistant_message in history:
-        context += f"User: {user_message}\nAssistant: {assistant_message}\n"
-    context += f"User: {message}\nAssistant:"
-
-    # Tokenize the input
-    inputs = tokenizer(context, return_tensors="pt").to("cuda")
-
-    # Generate a response
-    outputs = model.generate(
-        input_ids=inputs.input_ids,
-        max_new_tokens=max_tokens,
+    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,
-        do_sample=True
-    )
+    ):
+        token = message.choices[0].delta.content
+
+        response += token
+        yield response
 
-    # Decode and return the response
-    response = tokenizer.decode(outputs[:, inputs.input_ids.shape[-1]:][0], skip_special_tokens=True)
-    return response
 
-# Build the Gradio ChatInterface
+"""
+For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
+"""
 demo = gr.ChatInterface(
-    fn=respond,
+    respond,
     additional_inputs=[
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max Tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=1.5, step=0.1, label="Temperature"),
-        gr.Slider(minimum=0.1, maximum=1.0, value=0.9, step=0.05, label="Top-p"),
+        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)",
+        ),
     ],
 )
 
+
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()