Update app.py

app.py

@@ -3,7 +3,7 @@ os.environ["HF_HUB_ENABLE_HF_TRANSFER"] = "False"
 os.environ["TOKENIZERS_PARALLELISM"] = "true"
 import numpy as np
 import gradio as gr
-import spaces
+#import spaces
 import torch
 import torch.nn.functional as F
 from PIL import Image
@@ -34,7 +34,7 @@ model.eval()
 mask_token_id = model.config.mask_token_id
 
 
-@spaces.GPU
+#@spaces.GPU
 def text_to_image_generation(input_text, guidance_scale=1.75, generation_timesteps=18):
     prompts = [input_text]
     config.training.batch_size = config.batch_size = 1
@@ -92,7 +92,7 @@ def text_to_image_generation(input_text, guidance_scale=1.75, generation_timeste
     return images[0]
 
 
-@spaces.GPU
+#@spaces.GPU
 def text_guided_inpainting(input_text, inpainting_image, inpainting_mask_input, guidance_scale=1.75, generation_timesteps=16):
     # pre-process inpainting mask
     alpha_channel = inpainting_mask_input["layers"][0][:, :, 3]
@@ -176,7 +176,7 @@ def text_guided_inpainting(input_text, inpainting_image, inpainting_mask_input,
     return images[0]
 
 
-@spaces.GPU
+#@spaces.GPU
 def text_guided_extrapolation(input_img, input_text, left_ext, right_ext, guidance_scale=1.75, generation_timesteps=16):
     config.offset = 0
     config.training.batch_size = config.batch_size = 1
@@ -301,7 +301,7 @@ def text_guided_extrapolation(input_img, input_text, left_ext, right_ext, guidan
     return images[0]
 
 
-@spaces.GPU
+#@spaces.GPU
 def multimodal_understanding(input_img, input_text, chat_history):
     top_k = 1  # retain only the top_k most likely tokens, clamp others to have 0 probability
 
@@ -549,4 +549,4 @@ with gr.Blocks() as demo:
             clear = gr.ClearButton([msg, chatbot])
     msg.submit(multimodal_understanding, [chat_input_img, msg, chatbot], [msg, chatbot])
 
-demo.launch()
+demo.launch(share = True)