[Tracker] Nested tensor op coverage requests #118107
Description
This issue tracks op coverage requests for nested tensors. If you'd like a specific op to be implemented for nested tensors, please add a comment here so we can prioritize effectively.
Prior requests:
- Conv2d: Nested tensors fail on Conv2D #114090
- KV-cache related ops: Implement KV cached attention with NestedTensor #115978
- Includes ops that mutate in-place
- Slicing of NTs across constant dimension: Allow slicing of Nested Tensors along constant dimensions #108567
- EmbeddingBag: EmbeddingBag to support mini-batches with offsets #93843
- Loss functions
- Cross-entropy, log_softmax, NLL loss): More Nested Tensor Functionality (layer_norm, cross_entropy / log_softmax&nll_loss) #99142
- Binary cross-entropy: binary_cross_entropy/bce_with_logits (+ other loss functions) for nested_tensor #89125
eq,to,masked_select,index_select,narrow: Add eq, to, masked_select, index_select, narrow to nested tensors #88137- LayerNorm on non-contiguous (transposed) NJT: nested_tensor LayerNorm fails with shape (*,*,j1,*) (but works with (*,j1,*,*)) #130538