Update app.py

app.py

@@ -40,7 +40,7 @@ class HFEmbedder(nn.Module):
 
 def load_t5(device: str | torch.device = "cuda", max_length: int = 512) -> HFEmbedder:
     # max length 64, 128, 256 and 512 should work (if your sequence is short enough)
-    return HFEmbedder("google/t5-v1_1-xxl", max_length=max_length, torch_dtype=torch.bfloat16).to(device)
+    return HFEmbedder("city96/t5-v1_1-xxl-encoder-bf16", max_length=max_length, torch_dtype=torch.bfloat16).to(device)
 
 if __name__ == "__main__":
     device = "cuda" if torch.cuda.is_available() else "cpu"