-
Hi @yuxumin,
Thanks for your contributions to build new benchmarks ShapeNet55/34. I have a question about the shapenet dataloader part. In the dataloader, the pc_norm function will normalize the poin…
-
**Description**
I'm converting a pytorch model to torchscript with or without fp16 precision, and I get much slower triton inference when using FP16, even though a torchscript benchmark shows it to b…
-
Hi,
I have been looking at approximate activation functions and did some analysis on `ultra_fast_sigmoid` which is included in Theano.
I found that compared to some other sigmoid approximations…
-
### Motivation
When we use LMDeploy for Serving, although throughput is also a concern, **more emphasis is placed on throughput under latency constraints with different QPS**. This is a performance m…
-
This entrance task is about doing benchmarks against [sklearn v0.18](http://scikit-learn.org/dev/modules/gaussian_process.html), [gpy](https://github.com/SheffieldML/GPy), [gpflow](https://github.com/…
-
## ❓ Questions and Help
Hi @KaihuaTang
Thanks for the code for scene graph detection. I was trying to run the code for scene graph generation on custom images but the results I am getting are too…
-
This is a development requirement documentation of a feature we wish to collaborate with the community.
In this issue, we wish to welcome the community developers to participate in the development of…
-
### 🐛 Describe the bug
profile - https://fburl.com/scuba/pyperf_experimental/on_demand/ghc8lpjr
xref - https://fb.workplace.com/groups/1075192433118967/permalink/1425136311457909/
Repro - D570909…
-
## Problem
I have modified the tutorial code of `https://github.com/openai/triton/blob/main/python/tutorials/03-matrix-multiplication.py` to test int8 GEMM of Triton. Source code is as follows:
…
-
Fails on running harness
Traceback (most recent call last):
File "code/main.py", line 279, in handle_run_harness
result = harness.run_harness()
File "/mlperf/code/common/harness.py", lin…