liangzelang

Could you provide more detail info about this project

1

Thank you to share the OCR project, and I also have great interest in this project but I cannot find more detail about this project. 1. how to install &...

thank you for your great work .and I have used your open source to finish my project .but there are some questions whick I can't understand and solve. 1、I have...

I want run vulkan ops test on Android , so I build this project, but build failed. ``` mkdir build && cd build (ggml_py310) [root@engine-liangzelang-sicicc-0 build]$cmake .. -DCMAKE_SYSTEM_NAME=Android -DCMAKE_SYSTEM_VERSION=33 -DCMAKE_ANDROID_ARCH_ABI=arm64-v8a...

Gemv with half precision cannot get correct result.

2

Through [this Issue ](https://github.com/CNugteren/CLBlast/issues/560) I selected the expected high-performance operator, but found a problem. If the data type is float, the calculation result is correct, but if the data type...

Why clblast::Gemv API is slower than clblast_tune_xgemv when m = 2048 n=16384 precision=fp16.

10

Hi, I recently used CLblast to speed up Android devices, but found a problem. The performance of using the Gemv API cannot reach the performance of the tune program, and...

Fused_MHA does not support when seq_len = 1024, Dh==72, causal_mask==false

1

When I use FMHA_v2, I found it does not support my scenes. So i wonder is there any way to use fmha_v2 except changing model. Thx a lot.