Update app.py

app.py

@@ -56,7 +56,7 @@ print(HF_TOKEN)
 
 
 m = AutoModelForCausalLM.from_pretrained(
-    "google/gemma-2b-it", 
+    "ibm-granite/granite-guardian-3.0-2b", 
     torch_dtype=torch.bfloat16 if torch.cuda.is_available() else torch.float32, 
     trust_remote_code=True,token=HF_TOKEN
 )
@@ -64,7 +64,7 @@ m = AutoModelForCausalLM.from_pretrained(
 embedding_func=m.get_input_embeddings()
 embedding_func.weight.requires_grad=False
 
-tok = AutoTokenizer.from_pretrained("google/gemma-2b-it", 
+tok = AutoTokenizer.from_pretrained("ibm-granite/granite-guardian-3.0-2b", 
                                     trust_remote_code=True,token=HF_TOKEN
                                    )
 tok.padding_side = "left"
@@ -100,7 +100,7 @@ def embedding_shift(original_embedding,shift_embeddings,prefix_embedding,suffix_
         )
     return input_embeddings
 
-@spaces.GPU(duration=10)
+@spaces.GPU(duration=30)
 def engine(input_embeds):
     m.to("cuda")
     output_text = []