sitabulaixizawaluduo

[Performance]: VariadicSplit Op's CPU time is different between 2024.0.0 and 2023.0.0

5

### OpenVINO Version 2024.0.0 ### Operating System Ubuntu 22.04 (LTS) ### Device used for inference CPU ### OpenVINO installation Build from source ### Programming Language Python ### Hardware Architecture x86...

The ppl value of the opt-6.7b-smoothquant model shows abnormal performance

1

I tested the mit-han-lab/opt-6.7b-smoothquant model and the opt-6.7b model on HuggingFace. The ppl obtained using the WikiText-2 dataset was 20.65 and 10.92, respectively. The tests were conducted on an A30...

[Performance]:Why is there a reorder op after variadicsplit op ?

2

### OpenVINO Version 2024.0.0 ### Operating System Ubuntu 22.04 (LTS) ### Device used for inference CPU ### OpenVINO installation Build from source ### Programming Language Python ### Hardware Architecture x86...

Is there any plans to support Int8 weight quant ?

2

The int8 quantization is basically lossless at the moment, and although awq is good in terms of accuracy and performance, int8 is a better choice in some scenarios where accuracy...

[Bug] DeepSeek-R1 NCCL WatchDog Timeout Error

3

### Checklist - [x] 1. I have searched related issues but cannot get the expected help. - [x] 2. The bug has not been fixed in the latest version. -...