Update app.py

app.py

@@ -13,21 +13,14 @@ FINETUNED_MODEL = "LaibaIrfan/emoji_math"
 # Load tokenizer and model
 @lru_cache()
 def load_model():
-    tokenizer = AutoTokenizer.from_pretrained(FINETUNED_MODEL)
-
+    tokenizer = AutoTokenizer.from_pretrained(BASE_MODEL, trust_remote_code=True)  # Use base model tokenizer
     base_model = AutoModelForCausalLM.from_pretrained(
         BASE_MODEL,
-        torch_dtype=torch.float16,  # Use float16 for efficiency
-        device_map="auto",          # Auto-assign device (GPU if available)
-        load_in_8bit=True           # Reduce memory usage (slightly increases inference time)
-    )
-
-    model = PeftModel.from_pretrained(
-        base_model,
-        FINETUNED_MODEL,
-        device_map="auto" 
+        torch_dtype=torch.float16,
+        device_map="auto",
+        load_in_8bit=True
     )
-
+    model = PeftModel.from_pretrained(base_model, FINETUNED_MODEL, device_map="auto")
     return tokenizer, model
 
 # Load the model