-
The behavior of some operations, e.g. binary elementwise, is not obvious when mixing `Tensor`/`Shape` types. We may want to disallow mixing types in such cases. See https://github.com/NVIDIA/TensorRT-…
-
## List of Non-Deterministic Operations in PyTorch
The following operations in PyTorch exhibit non-deterministic behavior according to the ```torch.use_deterministic_algorithms``` [documentation](h…
-
It would be nice to have a function like `drop_block`, except that one would specify a selection/filter of blocks to keep.
Something like this would work:
```py
def filter_keys(tensor: TensorM…
-
I know quantized model isn't supported yet. I would like to confirm this is the symptom due to Dequantize/Quantize operators?
The model I'm testing is ShuffleNet-v2-int8 from [here](https://github.…
-
Hi @nullplay,
I wanted to start a discussion on Finch-MLIR MLIR tensors API. In https://github.com/pydata/sparse/tree/main/sparse/mlir_backend we have an initial Tensor class which provides constr…
-
Currently, our bitwise operators support tensor–scalar operations, namely
- `ttnn.bitwise_and`
- `ttnn.bitwise_or`
- `ttnn.bitwise_xor`
Extending the second argument (RHS) from a scalar to…
-
Hi. Something (possibly not unsloth) changed between July and now.
I am getting an unexpected OOM error trying to do a LORA finetune. This worked before, but is now barfing.
Looked at #338, but not…
-
**Describe the bug**
ttnn.max op throws `Unsupported dim` error while reducing along the `channel dimension` for 4d input and `batch dimension` for both 4d and 3d input.
The `Unsupported dim` iss…
-
Currently, [the code for tensoring a matrix](https://github.com/vprusso/toqito/blob/dd08a53e76d600532fb7e8ec0be494ad8a8b05ef/toqito/matrix_ops/tensor.py#L135C1-L146C22) with itself is the following:
…
-
I am trying to perform the following tensor operations [torch.cumsum()](https://pytorch.org/docs/stable/generated/torch.cumsum.html), and the quickest way i can think of is to convert the tensor to a …