Update app.py

app.py

@@ -12,10 +12,31 @@ def install_flashattn():
      subprocess.run(['sh', './flashattn.sh'])
 install_flashattn()
 
-import time
-import gradio as gr
+# --- PyTorch Environment Setup ---
+os.environ['PYTORCH_NVML_BASED_CUDA_CHECK'] = '1'
+os.environ['TORCH_LINALG_PREFER_CUSOLVER'] = '1'
+os.environ['PYTORCH_ALLOC_CONF'] = 'expandable_segments:True,pinned_use_background_threads:True'
+os.environ["SAFETENSORS_FAST_GPU"] = "1"
+os.environ['HF_HUB_ENABLE_HF_TRANSFER'] = '1'
+
 import torch as th
 import torch
+
+# Set precision settings for reproducibility and performance
+torch.backends.cuda.matmul.allow_tf32 = False
+torch.backends.cudnn.allow_tf32 = False
+torch.backends.cuda.matmul.allow_bf16_reduced_precision_reduction = False
+torch.backends.cuda.matmul.allow_fp16_reduced_precision_reduction = False
+torch.backends.cudnn.deterministic = False
+torch.backends.cudnn.benchmark = False # Set to True for potential speedup if input sizes are static, False for dynamic
+torch.backends.cuda.preferred_blas_library="cublas"
+torch.backends.cuda.preferred_linalg_library="cusolver"
+torch.set_float32_matmul_precision("highest")
+
+
+import time
+import gradio as gr
+
 import numpy as np
 import tempfile
 from diffusers import AutoencoderKLWan