rwkv.cpp icon indicating copy to clipboard operation
rwkv.cpp copied to clipboard

INT4/INT5/INT8 and FP16 inference on CPU for RWKV language model

Results 40 rwkv.cpp issues
Sort by recently updated
recently updated
newest added

基本支持RWKV6 转换,加载,lora好了,但是计算图有问题,rwkv_graph.inc:348行附近怎么都改不好,我打了感叹号标记

Needs https://github.com/RWKV/ggml/pull/3

Hello, thanks for the project, but it is hard to get it running at first. Could you just please provide some simple Colab notebook? That is much better than setup...

Is there support for RWKV6?

https://docs.github.com/en/actions/using-github-hosted-runners/about-github-hosted-runners/about-github-hosted-runners This helps resolve: https://github.com/saharNooby/rwkv.cpp/issues/162 (if it fully works, not fully tested)

Since apple silicon is now supported on github https://twitter.com/simonw/status/1752068065855148291

Because instruction for Apple silicon in Readme.md has been applied, CMAKE_SYSTEM_PROCESSOR from `cmake .` appears arm64. But, there is warning as below for `cmake .` ``` -- Accelerate framework found...

我修改了 `CMakeLists.txt` 添加安装 `librwkv.so` 的功能( `make install/cmake --install` ), 并且支持 `pkg-config`。 使用效果: ```bash meng@laptop:/tmp$ cat main.c #include "rwkv.h" #include int main(void) { printf("%s\n", rwkv_get_system_info_string()); return 0; } meng@laptop:/tmp$ echo gcc...

# Add API server ## start ``` python rwkv/api.py RWKV-4-World-3B-v1-20230619-ctx4096_Q5_1.bin world 0.0.0.0 8000 ``` ## /completions ``` curl localhost:8000/v1/completions -X POST -H 'Content-Type: application/json' -d '{"prompt": "你是一个历史老师,请告诉我周文王是谁?", "stream": false}' -->...