inital deployment1

app.py

@@ -3,7 +3,7 @@ import cv2
 import torch
 import numpy as np
 from diffusers import StableDiffusionPipeline
-from transformers import AutoProcessor, AutoModel
+from transformers import AutoProcessor, AutoModel, AutoTokenizer
 from PIL import Image
 
 # Load the Real-Time Latent Consistency Model
@@ -13,6 +13,7 @@ device = "cuda" if torch.cuda.is_available() else "cpu"
 # Load the model (optimized for inference)
 model_id = "https://huggingface.co/spaces/radames/Real-Time-Latent-Consistency-Model"
 
+tokenizer = AutoTokenizer.from_pretrained(model_id)
 
 realtime_pipe = StableDiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float16)
 realtime_pipe.to("cuda")  # Use GPU for faster inference
@@ -52,4 +53,4 @@ with gr.Blocks() as demo:
     prompt_input = gr.Textbox(label="Real-Time Latent Consistency Model Prompt", value="A futuristic landscape")
     webcam_feed.change(fn=video_stream, inputs=[prompt_input], outputs=[processed_image, canvas])
 
-demo.launch()
+demo.launch(share=True)