Update app.py

app.py

@@ -5,10 +5,20 @@ import re
 import spaces
 import gradio as gr
 import torch
-from transformers import AutoTokenizer, AutoModelForCausalLM
+# from transformers import AutoTokenizer, AutoModelForCausalLM
 
-tokenizer = AutoTokenizer.from_pretrained("FlawedLLM/Bhashini_gemma_merged4bit_clean_final")
-model = AutoModelForCausalLM.from_pretrained("FlawedLLM/Bhashini_gemma_merged4bit_clean_final", load_in_4bit=True, device_map="auto")
+# tokenizer = AutoTokenizer.from_pretrained("FlawedLLM/Bhashini_gemma_merged4bit_clean_final")
+# model = AutoModelForCausalLM.from_pretrained("FlawedLLM/Bhashini_gemma_merged4bit_clean_final", load_in_4bit=True, device_map="auto")
+from transformers import AutoModelForCausalLM, AutoTokenizer
+from peft import PeftModel
+
+# 1. Load Your Base Model and LoRA Adapter
+model_name_or_path = "FlawedLLM/Bhashini_gemma_merged4bit_clean_final"  # Hugging Face model or local path
+lora_weights = "FlawedLLM/Bhashini_gemma_lora_clean_final"  # LoRA weights
+
+tokenizer = AutoTokenizer.from_pretrained(model_name_or_path)
+model = AutoModelForCausalLM.from_pretrained(model_name_or_path, load_in_8bit=True, device_map='auto')
+model = PeftModel.from_pretrained(model, lora_weights)
 
 # alpaca_prompt = You MUST copy from above!
 @spaces.GPU(duration=300)