fix cache again

app.py

@@ -154,6 +154,7 @@ def bot_response(history, model_selection, max_tokens, temperature, top_k, top_p
             tokenizer = AutoTokenizer.from_pretrained(MODELS[model_name]["base_model"])
             model = AutoModelForCausalLM.from_pretrained(**load_kwargs).to("cuda")
             reward_model = RewardModel(model_name=MODELS[model_name]["reward_repo_id"])
+            CACHE["model_name"] = model_name
             CACHE["tokenizer"] = tokenizer
             CACHE["model"] = model
             CACHE["reward_model"] = reward_model