Update app.py

app.py

@@ -13,6 +13,7 @@ from kolors.models.modeling_chatglm import ChatGLMModel
 from kolors.models.tokenization_chatglm import ChatGLMTokenizer
 from diffusers import UNet2DConditionModel, AutoencoderKL
 from diffusers import EulerDiscreteScheduler
+import subprocess
 
 # Initialize models
 device = "cuda" if torch.cuda.is_available() else "cpu"
@@ -21,6 +22,8 @@ dtype = torch.float16
 # Download Kolors model
 ckpt_dir = snapshot_download(repo_id="Kwai-Kolors/Kolors")
 
+subprocess.run('pip install flash-attn --no-build-isolation', shell=True)
+
 # Load Kolors models
 text_encoder = ChatGLMModel.from_pretrained(os.path.join(ckpt_dir, 'text_encoder'), torch_dtype=dtype).to(device)
 tokenizer = ChatGLMTokenizer.from_pretrained(os.path.join(ckpt_dir, 'text_encoder'))