-
### š Describe the bug
torchbench_amp_fp16_training
xpu train moco
Traceback (most recent call last):
File "/home/sdp/actions-runner/_work/torch-xpu-ops/pytorch/benchmarks/dynamo/common.py", liā¦
-
https://www.npmjs.org/package/benchmark
Almost certain to be less wonky than the one-off thing I wrote.
-
## š Bug
Running the upstreamed benchmarking scripts with the following command results in an unexpected error.
```bash
python xla/benchmarks/experiment_runner.py \
--suite-name torchbeā¦
-
## Problem
Running [Noir's `bench-acir-bb`](https://github.com/AztecProtocol/aztec-packages/blob/0e8e772c05c81c0b5cf6a2d047a5587c6c2e1a9c/noir/Earthfile#L233-L237) Earthly target currently compilesā¦
-
Following up from https://discourse.llvm.org/t/rfc-what-are-the-blockers-to-turning-on-gvnsink-by-default/72326. Before asking for review we should measure the impact of this optimization at various oā¦
-
Currently the CI benchmarks use a Lithops runtime image that I've built for the latest release of Cubed. We should build the images as a part of the CI workflow.
-
### Checklist
- [X] I added a descriptive title
- [X] I searched open reports and couldn't find a duplicate
### Summary
This project has not been touched for some time; therefore, the goal oā¦
-
I'm putting this note here though it could just as easily go [here](https://github.com/ProjectSidewalk/sidewalk-cv-2021) or [here](https://github.com/ProjectSidewalk/sidewalk-cv-tools). It's just thatā¦
-
For an Open submission, we have tried setting TTFT/TPOT latencies in `user.conf` e.g.:
mixtral-8x7b.Server.ttft_latency = 3000
mixtral-8x7b.Server.tpot_latency = 300
However, in the resultingā¦
-
We have made the choice of full and modified Newton method optional.
Full Newton method is still the default since FV is also still the default, for which the modified Newton showed mixed results witā¦