init

app.py

@@ -39,7 +39,7 @@ config.encoder_path = hf_hub_download("csuhan/TA-Tok", "ta_tok.pth")
 config.decoder_path = hf_hub_download("peizesun/llamagen_t2i", "vq_ds16_t2i.pt")
 inference = TextToImageInference(config)
 
-@spaces.GPU(duration=240)
+@spaces.GPU(duration=120)
 def generate_image(prompt, resolution, top_p, top_k, cfg_scale):
     image = inference.generate_image(prompt, resolution, top_p, top_k, cfg_scale)
     return image
@@ -47,7 +47,7 @@ def generate_image(prompt, resolution, top_p, top_k, cfg_scale):
 def clear_inputs_t2i():
     return "", None
 
-@spaces.GPU(duration=240)
+@spaces.GPU(duration=120)
 def understand_image(image, prompt):
     return generate_text(inference, image, prompt)