Replace cache_resource with cache

app.py

@@ -26,7 +26,7 @@ def typical_sampling(model, input_ids, attention_mask, no_repeat_ngram_size, max
     )
 
 
-@st.cache_resource
+@st.cache(allow_output_mutation=True)
 def setModel():
     model = AutoModelForCausalLM.from_pretrained(MODEL)
     tokenizer = AutoTokenizer.from_pretrained(TOKENIZER)