inference

inference copied to clipboard

Bump transformers from 4.33.2 to 4.36.0 in /text_to_image

3

Bumps [transformers](https://github.com/huggingface/transformers) from 4.33.2 to 4.36.0. Release notes Sourced from transformers's releases. v4.36: Mixtral, Llava/BakLlava, SeamlessM4T v2, AMD ROCm, F.sdpa wide-spread support New model additions Mixtral Mixtral is the new...

dependabot[bot]

[Stable Diffusion Reference]: `qps` can't set to float

2

I run and have an error: ``` time python3 main.py --dataset "coco-1024" --dataset-path coco2014_full --profile stable-diffusion-xl-pytorch --model-path model/stable_diffusion _fp16/ --dtype fp16 --device cuda --scenario SingleStream --model-name stable-diffusion-xl --qps 0.022 --output...

maria-18-git

Stable Diffusion Reference support `qps` as input parameter in a command but don't support `target_latency` for SingleStream. @pgmpablo157321 could you please add this support ?

maria-18-git

**Current progress:** * CM coverage to automate and reproduce MLPerf inference: [GitHub](https://github.com/mlcommons/ck/issues/1052) * All reference implementations are supported by CM including GPT-J and Stable Diffusion (though didn’t run LLAMA). *...

gfursin

Does the tensorflow version of DLRM support cpus?

1

Does the tensorflow version of DLRM support cpus?

Bi9Boss

Stable Diffusion reference implementation is giving error on the accuracy run

5

Please see below for the detailed output. The run is done on Nvidia RTX 4090 GPU. ``` CMD: /home/arjun/cm/bin/python3 main.py --scenario SingleStream --profile stable-diffusion-xl-pytorch --dataset coco-1024 --dataset-path /home/arjun/CM/repos/local/cache/03fbdcf95b3d4104/install --dtype fp16...

arjunsuresh

Hi, I was running dlrm pytorch with CPU docker by using fake data. seeing below error. /root/mlcommons/recommendation/dlrm/pytorch/python/dlrm_data_pytorch.py:328: UserWarning: Creating a tensor from a list of numpy.ndarrays is extremely slow. Please...

sivanaga

multicard support and data distribution across ranks missing in the reference for SDXL (https://github.com/mlcommons/inference/tree/master/text_to_image). @pgmpablo157321 can you add this support ?

kumarans-ai

[root@poc4 bert]# make setup make[1]: Entering directory '/home/sunyu/benchmark/MLPerf/inference/language/bert' You need to run this command from the toplevel of the working tree. make[1]: *** [init_submodule] Error 1 make[1]: Leaving directory '/home/sunyu/benchmark/MLPerf/inference/language/bert'...

Yunhui1

Is it possible to calculate the LLM performance, Token per Second ?

1

Hello, I get the inference result v3.1 and analysis the performance, throughput (token per second). In the large language model task, the test results are measured by Queries/s and Samples/s....

ConstantPark