Update app.py

app.py

@@ -25,9 +25,12 @@ model_id = "FlawedLLM/BhashiniLLM"
 
 # Load the base model (the one you fine-tuned with LoRA)
 base_model = AutoModelForCausalLM.from_pretrained(model_id, device_map='auto')  # Load in 8-bit for efficiency
+for param in base_model.parameters():
+    param.data = param.data.to(torch.float16)  # or torch.float32
 
 # Load the LoRA adapter weights
 model = PeftModel.from_pretrained(base_model, model_id)
+
 tokenizer = AutoTokenizer.from_pretrained(model_id)