BackendCompilerFailed

Running modernbert within transformer Trainer class, with torch_compile=False, I get the following error:

```
raise AssertionError(
torch._dynamo.exc.BackendCompilerFailed: backend='inductor' raised:
AssertionError: Please convert all Tensors to FakeTensors first or instantiate FakeTensorMode with 'allow_non_fake_inputs'. Found in aten.embedding.default(tensor([...], device='cuda:0', size=(50368, 768), grad_fn=), FakeTensor(..., device='cuda:0', size=(1, s1), dtype=torch.int64), 50283)
```

torch version: '2.4.1+cu121'
transformers version: '4.48.3'

I am not sure what the problem is here.

No project files or tests are named. Start by reproducing the ModernBERT Trainer run with torch 2.4.1+cu121 and transformers 4.48.3, recording whether torch_compile is actually disabled; done means the reported BackendCompilerFailed error is isolated to a reproducible project path or resolved with a verified regression test.

Written by the indexing model from the issue text.