Update app.py

app.py

@@ -1,7 +1,6 @@
 import re
 import spaces
 import gradio as gr
-from transformers import AutoTokenizer, AutoModel, BitsAndBytesConfig
 import torch
 # from peft import PeftModel, PeftConfig
 
@@ -43,13 +42,11 @@ import torch
 #         load_in_4bit = True,
 #     )
 # tokenizer = AutoTokenizer.from_pretrained("FlawedLLM/Bhashini")
-from unsloth import FastLanguageModel
-model, tokenizer = FastLanguageModel.from_pretrained(
-        model_name = "FlawedLLM/Bhashini_2", # YOUR MODEL YOU USED FOR TRAINING
-        max_seq_length = 2048,
-        dtype = torch.float16,
-        load_in_4bit = True,)
-FastLanguageModel.for_inference(model)
+# Load model directly
+from transformers import AutoTokenizer, AutoModelForCausalLM
+
+tokenizer = AutoTokenizer.from_pretrained("FlawedLLM/Bhashini_9")
+model = AutoModelForCausalLM.from_pretrained("FlawedLLM/Bhashini_9", device_map='auto')
 
 @spaces.GPU(duration=300)
 def chunk_it(input_command):