Update app.py

app.py

@@ -101,7 +101,7 @@ def embedding_shift(original_embedding,shift_embeddings,prefix_embedding,suffix_
         )
     return input_embeddings
 
-@spaces.GPU
+@spaces.GPU(duration=10)
 def engine(input_embeds):
     m.to("cuda")
     output_text = []
@@ -120,14 +120,14 @@ def engine(input_embeds):
             output_text += tok.batch_decode(outputs, skip_special_tokens=True)
     return output_text
     
-@spaces.GPU
+@spaces.GPU(duration=10)
 def chat_engine(input_ids):
     m.to("cuda")
     prompt_length=len(input_ids[0])
     with torch.no_grad():
         outputs = m.generate(
             input_ids = input_ids.to("cuda"),
-            max_new_tokens = 512,
+            max_new_tokens = 16,
             do_sample = True,
             temperature = 0.6,
             top_p = 0.9,