update/modified app.py

app.py

@@ -1,12 +1,20 @@
 import gradio as gr
-from huggingface_hub import InferenceClient
+from transformers import AutoModelForCausalLM, AutoTokenizer
+from peft import PeftModel
 
-"""
-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("sreyanghosh/lora_model")
+# Load the model and tokenizer
+def load_model():
+    base_model_name = "unsloth/llama-3.2-1b-instruct-bnb-4bit"  # Replace with your base model name
+    lora_model_name = "sreyanghosh/lora_model"  # Replace with your LoRA model path
+    tokenizer = AutoTokenizer.from_pretrained(base_model_name)
+    model = AutoModelForCausalLM.from_pretrained(base_model_name, device_map="auto")
+    model = PeftModel.from_pretrained(model, lora_model_name)
+    model.eval()
+    return tokenizer, model
 
+tokenizer, model = load_model()
 
+# Define the respond function
 def respond(
     message,
     history: list[tuple[str, str]],
@@ -15,34 +23,36 @@ def respond(
     temperature,
     top_p,
 ):
+    # Prepare the conversation history
     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]})
-
+    for user_input, bot_response in history:
+        if user_input:
+            messages.append({"role": "user", "content": user_input})
+        if bot_response:
+            messages.append({"role": "assistant", "content": bot_response})
     messages.append({"role": "user", "content": message})
 
-    response = ""
-
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
+    # Format the input for the model
+    conversation_text = "\n".join(
+        f"{msg['role']}: {msg['content']}" for msg in messages
+    )
+    inputs = tokenizer(conversation_text, return_tensors="pt", truncation=True)
+    
+    # Generate the model's response
+    outputs = model.generate(
+        inputs.input_ids,
+        max_length=len(inputs.input_ids[0]) + max_tokens,
         temperature=temperature,
         top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
+        pad_token_id=tokenizer.eos_token_id,
+    )
+    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
 
-        response += token
-        yield response
+    # Extract the new response
+    new_response = response[len(conversation_text):].strip()
+    yield new_response
 
-
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
+# Gradio app configuration
 demo = gr.ChatInterface(
     respond,
     additional_inputs=[
@@ -59,6 +69,5 @@ demo = gr.ChatInterface(
     ],
 )
 
-
 if __name__ == "__main__":
     demo.launch()

requirements.txt

@@ -1,3 +1,5 @@
 huggingface_hub==0.25.2
 gradio
-unsloth
+transformers
+peft
+torch