-
Hi @willow-ahrens @hameerabbasi,
I managed to run first simple benchmarks in `pydata/sparse` with `PyData` and `Finch` backends.
[`asv`](https://asv.readthedocs.io/en/v0.6.1/index.html) tool was u…
-
Bug description:
When converting a cuda tensor, which is indexed from an other large tensor to CPU memory scalar, the time consumption of this process is unusually long.
Minimal reproducible examp…
-
hello, when run main_finetune.py till 238th row:
for param in fsdp_ignored_parameters:
dist.broadcast(param.data, src=dist.get_global_rank(fs_init.get_data_parallel_group(), 0),
…
-
Right now what we have is docstrings but they could use work - this came up as @vayuda was looking at extending his bitpacking work to include a notion of scales
1. What does tensor core layout mea…
-
In the BuildArmComputeTensor(Tensor& tensor, const armnn::TensorInfo& tensorInfo) function in the ArmComputeTensorUtils.hpp file, I tried to modify the alignment value of ComputeTensor memory, that is…
-
I've downloaded the latest version of the repo using
`pip install git+https://github.com/microsoft/fadtk.git`
However, when I go to run the command:
`fadtk vggish dataset/one/ dataset/two/`
…
-
### Issue type
Bug
### Have you reproduced the bug with TensorFlow Nightly?
Yes
### Source
binary
### TensorFlow version
v2.17.0
### Custom code
No
### OS platform and distribution
Ubuntu 2…
-
Would love a way to print out `tensor`s, in a similar way to `vector.print`.
Seems like the alternative (relatively simple way) would be to first lower to memref (after talking to @makslevental), …
-
The input is
```python
import dpctl.tensor as dpt
val = dpt.zeros(2, dtype="u1")
max = dpt.zeros(tuple(), dtype="i1")
dpt.clip(val, min=None, max=max)
```
Raises `ValueError: function 'cli…
-
~/projects/piper/piper/src/python_run $ python3 -m piper.http_server --model ../../../voices/german/de_DE-thorsten-high.onnx
* Serving Flask app 'http_server'
* Debug mode: off
INFO:werkzeug:WARN…