Replace torch.jit.instance with instance to assist torch.compile (#522)

Summary:

torch.jit.instance is mainly needed to refine Generics, but in this case instance is sufficient, instance unlike torch.jit.instance is also playing nicely with torch.compile

Differential Revision: D53797660

torchmultimodal/modules/layers/transformer.py

@@ -381,9 +381,8 @@ def _cross_attention_block(
  # TODO: figure out caching for cross-attention
  use_cache=False,
  )
- assert torch.jit.isinstance(
- output, Tensor
- ), "cross-attention output must be Tensor."
+ assert isinstance(output, Tensor), "cross-attention output must be Tensor."
+
  attention_output = self.cross_attention_dropout(output)
  return attention_output