Qwen-Image-to-LoRA

Running on Zero

Alexander Bagus commited on 2 days ago

Commit

507a133

1 Parent(s): ec96f0a

22

Files changed (1) hide show

app.py CHANGED Viewed

@@ -62,7 +62,7 @@ vram_config = {
     "offload_dtype": "disk",
     "offload_device": "disk",
     "onload_dtype": torch.bfloat16,
-    "onload_device": "gpu",
     "preparing_dtype": torch.bfloat16,
     "preparing_device": "cuda",
     "computation_dtype": torch.bfloat16,

     "offload_dtype": "disk",
     "offload_device": "disk",
     "onload_dtype": torch.bfloat16,
+    "onload_device": "cuda",
     "preparing_dtype": torch.bfloat16,
     "preparing_device": "cuda",
     "computation_dtype": torch.bfloat16,