blob: f193130eea05f84550c3fce96f591cfbb90da1be [file] [log] [blame] [view]
# Quantized Ops
This folder contains kernels for quantization related ops, similar to `//executorch/kernels/portable/`.
## How to add quantized ops
1. Add a new operator definition in `quantized.yaml`
2. Implement the kernel for this operator.
3. Add unit test in `/test` directory.
4. Start to use `//executorch/kernels/quantized:generated_lib` in ExecuTorch or
create your own generated lib if you only need a subset of the ops.