Update app.py

app.py

@@ -10,16 +10,14 @@ device = "cpu"
 model_repo_id = "stabilityai/sdxl-turbo"  # Continuando com o modelo especificado
 
 # Carregar o pipeline configurado para CPU
-pipe = DiffusionPipeline.from_pretrained(model_repo_id,revision="fp16",)
+pipe = DiffusionPipeline.from_pretrained(model_repo_id)
 pipe = pipe.to(device)
 
-# Atualize as configurações do pipeline diretamente
-pipe.unet.config.sample_size = (808, 512)  # Ajuste de altura e largura divisíveis por 8
-
 # Carregando o dataset do Hugging Face
 dataset = load_dataset("LEIDIA/Data_Womleimg")
 
 
+
 # Parâmetros para carregar o dataset personalizado
 dataset_descriptions = [
     "Images/Image01.jpg,A blonde woman wearing a blue leather jumpsuit with leather gloves the same color. She is standing with one hand on her face and the other on her waist, in a confident pose.  The background is dark and neutral, highlighting the woman figure.",
@@ -57,6 +55,8 @@ dataset_descriptions = [
     "Images/Image33.jpg,a woman on an urban street.  She is wearing a red leather coat open, revealing a top that exposes her abdomen, and a short dark red leather shorts. Her hair is long and loose, and she is wearing red lipstick. The background shows a busy street with blurred shops and people."
 ]
 
+
+    
 # Definir parâmetros padrão para geração rápida
 DEFAULT_PROMPT = "A beautiful brunette woman wearing a blue leather pants "
 DEFAULT_INFERENCE_STEPS = 6
@@ -70,13 +70,6 @@ def resize_to_divisible_by_8(image):
     new_height = height + (8 - height % 8) if height % 8 != 0 else height
     return image.resize((new_width, new_height))
 
-def infer_simple(prompt):
-    # Garantir que as dimensões sejam divisíveis por 8 antes de gerar
-    adjusted_height = IMAGE_HEIGHT - (IMAGE_HEIGHT % 8)
-    adjusted_width = IMAGE_WIDTH - (IMAGE_WIDTH % 8)
-
-    print(f"Using height: {IMAGE_HEIGHT} and width: {IMAGE_WIDTH}")
-
 # Função simples para gerar imagem
 def infer_simple(prompt):
     # Geração da imagem
@@ -111,8 +104,5 @@ with gr.Blocks() as demo:
         outputs=result,
     )
 
-result = gr.Image(label="Generated Image", shape=(808, 512))
-
-
 if __name__ == "__main__":
     demo.launch()