Update app.py

app.py

@@ -1,16 +1,23 @@
 # app.py
 
 import gradio as gr
-from transformers import pipeline
+from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
 
-# Load Falcon-7B-Instruct or fallback model
 model_id = "tiiuae/falcon-7b-instruct"
+
+# Load tokenizer and model with trust_remote_code
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+model = AutoModelForCausalLM.from_pretrained(model_id,
+                                             trust_remote_code=True,
+                                             torch_dtype="auto")
+
+# Create pipeline (no use_cache to avoid past_key_values crash)
 polisher = pipeline("text-generation",
-                    model=model_id,
-                    torch_dtype="auto",
-                    trust_remote_code=True)
+                    model=model,
+                    tokenizer=tokenizer,
+                    device=-1)
 
-def oxford_polish(sentence):
+def oxford_polish_strict(sentence: str) -> str:
     prompt = (
         "You are an Oxford grammar professor. "
         "Rewrite the following sentence in formal written English, following the Oxford University Style Guide. "