-
This issue is the same error as #8227, but it is not fixed by #8234.
I am running https://github.com/halide/Halide/commit/340136fec6d3ebc73e7a19eba1663e9b0ba8ab2d -- which includes #8234. Unlike #…
-
### What happened?
When compiling CLIP Text Encoder 1 ([stable_diffusion_xl_base_1_0_64_fp16_clip_1.mlir](https://storage.googleapis.com/shark-public/ean/sdxl-turbine/stable_diffusion_xl_base_1_0_64_…
-
I had the following expression in a calculation:
```
dot[B[ua]] -> IfB[+ dotXt[ua]
+ IfThen[fixAdvectionTerms!=0,
Upwind[beta[ub], Xt[ua], lb]…
-
Hey together,
I am having issues exporting a pytorch model to onnx via torch.onnx.dynamo_export.
I created a small code example for reproduction of the issue and have the following questions:
1. …
-
## 🚀 Feature
Currently you cannot use digits in names of tensors. Please allow digits, so that `torch.randn(64, 64, 4, 4, names=("height", "width", "channel_0", "channel_1"))` works rather than thr…
-
### Bug description
A fatal error occurs when decoding with a model that was converted to `packed8avx512` GEMM type from a .npz formatted model using `marian-conv`. I have a few models where this hap…
-
created from #7.
I don't know what original Triton code looked like that created this, but there is a division in the address expression.
[repros.zip](https://github.com/microsoft/triton-share…
-
Is the way dREL is formatted in `_method.expression` important? Some of them are hurting my eyes!
What's with all the whitespace between things?
```
With a as atom_site
label = a.la…
-
**Short description**
Probably just like #1093 but for encoding. We have a lot of data stored in `tfds.features.Sequence(tf.int64)` and `tfds.features.Sequence(tf.float32)` and the more data we add t…
-
Some PyTorch operations, like max and svd, can return a (named)tuple of tensors. There are two issues with their behavior:
- they do not always return a tuple
- when they do return a tuple, they c…