reduce gpu duration and add numpy and gradio dependencies

app.py

@@ -147,7 +147,7 @@ def load_model():
     model.eval()
     return model
 
-@spaces.GPU(duration=120)
+@spaces.GPU
 def gradio_interface_function(template, text, is_example):
     try:
         if len(tokenizer.tokenize(text)) > MAX_INPUT_SIZE:

requirements.txt

@@ -1,4 +1,6 @@
 transformers==4.41.2
 torch==2.2.2
 accelerate
-spaces>=0.1.0
+spaces>=0.1.0
+numpy<2.0
+gradio>=4.0.0