Fixed an issue with unsafe dimension expansion when handling 1D timestep tensors

[thkkk]

Added view(-1, 1) before expanding 1D tensors:

        # time embeddings
        if t.dim() == 1:
            t = t.expand(t.size(0), seq_len)

The original code attempted to directly expand 1D tensors to 2D, which could cause runtime errors since expand() only works on dimensions of size 1 (or just keep the size of dimensions). The fix ensures we first reshape 1D tensors to have a size-1 dimension before expansion.

        if t.dim() == 1:
            # Note: expand only works for dimensions of size 1; view(-1, 1) must be called first.
            t = t.view(-1, 1).expand(t.size(0), seq_len)
        elif t.dim() == 2 and t.size(1) == 1:
            t = t.expand(t.size(0), seq_len)

Moreover:

• Verified the fix handles various tensor shapes correctly
• The code correctness was validated via comprehensive i2v training experiments