-
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…
-
Add `pub use mlir_sys;`, since `mlir_sys` types appear in the public api.
`Inkwell` re-exports `llvm_sys` for the same reason.
https://www.lurklurk.org/effective-rust/re-export.html
-
```
module @outs_t27_4 {
func.func @main() -> tensor
-
I installed torch-mlir with following script in my own conda env, where 20240825.194 is the newest torch-mlir in the url [https://github.com/llvm/torch-mlir-release/releases/expanded_assets/dev-wheels…
-
Running tests with latest tt-mlir fails with the following errors:
```
2024-09-18 09:06:21.033 | FATAL | TTDevice - Tensor 0 - data type mismatch: expected UInt8, got UInt32
2024-09-1…
-
After the commit https://github.com/openxla/xla/commit/cb304cf640723620d9a67173e6bb650e34217169, JAX crashes in Triton on H100 with the following repro:
```python
import jax
import jax.numpy as j…
-
The root issue seems to be the conversion from a 0-sized traced array to a float. Is it supposed to be possible?
```julia
julia> using Reactant
julia> f(x) = vcat(x, x)
f (generic function wit…
-
## Description
**Job name:** [test_transpose[params5]](https://github.com/tenstorrent/tt-forge-fe/actions/runs/10956249916/job/30423853615)
**Regression commit:** 19732d0b2a10b053109d41243a29f95…
-
Source:
```cpp
#include
int main() {
std::cout
-
I have compiled my example by such pass pipeline.
oec-opt --stencil-shape-inference --convert-stencil-to-std --cse --parallel-loop-tiling='parallel-loop-tile-sizes=128,1,1' --canonicalize --test-gpu-…