## Linux

### 1. Requirements
- Already installed `CuDNN 9.x` & `CUDA 12.x` for _ONNX Runtime 1.23.2_

### 2. Inference
- `LD_LIBRARY_PATH=. ./asr --help`

- `LD_LIBRARY_PATH=. ./asr openai`

- `libonnxruntime_providers_cuda.so` & `libonnxruntime_providers_shared.so` - only need for CUDA inference

### 3. Build on Glibc>=2.28
- ``` sh
    mkdir -p ~/lib/onnxruntime
    cd ~/lib/onnxruntime
    wget https://github.com/microsoft/onnxruntime/releases/download/v1.23.2/onnxruntime-linux-x64-gpu-1.23.2.tgz
    tar -xzf onnxruntime-linux-x64-gpu-1.23.2.tgz
    export ORT_STRATEGY=manual
    export ORT_LIB_LOCATION=$HOME/lib/onnxruntime/onnxruntime-linux-x64-gpu-1.23.2/lib
    export LD_LIBRARY_PATH=$ORT_LIB_LOCATION:$LD_LIBRARY_PATH

    cargo clean
    cargo build --release -p asr --features cuda
```