-
### 🐛 Describe the bug
torchbench_amp_bf16_training
xpu train doctr_det_predictor
Traceback (most recent call last):
File "/home/sdp/actions-runner/_work/torch-xpu-ops/pytorch/benc…
-
### 🐛 Describe the bug
torchbench_amp_bf16_training
xpu train doctr_reco_predictor
Traceback (most recent call last):
File "/home/sdp/actions-runner/_work/torch-xpu-ops/pytorch/benc…
-
We should integrate a test suite(s) that checks for the performance of specific request scenarios.
One idea could be to have a slightly bigger test pbf that gets built regularly on master, like once …
-
Hi!
Example satellite data for testing purposes is provided inside the full MiRS distribution, mirs_v11r10_r111024075_20240315_code.tar.gz. This data lives in data/ExternalData/rdr/OrbitalMode/, an…
-
It would be nice to have some unit tests on the external interface of Kay so different optimizations can be shown to not break anything, and benchmarks can be added to see their effectiveness.
-
Thanks for your wonderful work!
But, as you mentioned: **"currently it only supports VOS datasets where all objects to track already appear on frame 0 in each video"**, which makes lots of benchmarks…
-
```
#!watchflakes
default 127.0.0.1:63484: use of closed network connection
Read 0, wanted 51, got , wanted 140303000101160303002800000000000000005df832a2c7d4ac17570d9fc1962a774871f425eccd…
-
Especially for server/client application benchmarking it would be helpful to be able to execute the client or server code-path in a non-monitored way.
As of now could not find a way to handle that …
-
**Is your feature request related to a problem? Please describe.**
Benchmarking certain FHIR Features in instrumented tests and/or benchmark modules is quite difficult. The team could use an app to c…
-
This is not an issue, just wanted to report that the conda version of this plugin along with the StreamHPC/openmm fork are working without issue on ROCm 5.0.2 and the the latest stable kernel(5.16.15)…