Remove examples/activation

This has become a standalone kernel in kernels-community. It also
precedes the relu example, which is much more minimal and easier to
understand.

Author: danieldk

.github/workflows/build_kernel.yaml

@@ -23,21 +23,16 @@ jobs:
           #authToken: "${{ secrets.CACHIX_AUTH_TOKEN }}"
         env:
           USER: github_runner
-      - name: Build activation kernel
-        run: ( cd examples/activation && nix build .\#redistributable.torch27-cxx11-cu126-x86_64-linux )
-      - name: Copy activation kernel
-        run: cp -rL examples/activation/result activation-kernel
+      - name: Build relu kernel
+        run: ( cd examples/relu && nix build .\#redistributable.torch27-cxx11-cu126-x86_64-linux )
+      - name: Copy relu kernel
+        run: cp -rL examples/relu/result relu-kernel
 
       - name: Build cutlass GEMM kernel
         run: ( cd examples/cutlass-gemm && nix build .\#redistributable.torch27-cxx11-cu126-x86_64-linux )
       - name: Copy cutlass GEMM kernel
         run: cp -rL examples/cutlass-gemm/result cutlass-gemm-kernel
 
-      - name: Build relu kernel
-        run: ( cd examples/relu && nix build .\#redistributable.torch27-cxx11-cu126-x86_64-linux )
-      - name: Copy relu kernel
-        run: cp -rL examples/relu/result relu-kernel
-
       - name: Build relu-backprop-compile kernel
         run: ( cd examples/relu-backprop-compile && nix build .\#redistributable.torch27-cxx11-cu126-x86_64-linux )
       - name: Copy relu-backprop-compile kernel

README.md

@@ -37,7 +37,7 @@ nix run nixpkgs#cachix -- use huggingface
 Then quick start a build with:
 
 ```bash
-cd examples/activation
+cd examples/relu
 nix run .#build-and-copy \
   --override-input kernel-builder github:huggingface/kernel-builder \
   --max-jobs 8 \
@@ -50,7 +50,7 @@ We also provide Docker containers for CI builds. For a quick build:
 
 ```bash
 # Using the prebuilt container
-cd examples/activation
+cd examples/relu
 docker run --rm \
   --mount type=bind,source=$(pwd),target=/kernelcode \
   -w /kernelcode ghcr.io/huggingface/kernel-builder:main build

docs/docker.md

@@ -28,8 +28,8 @@ installed.
 We provide a Docker image with which you can build a kernel:
 
 ```bash
-# navigate to the activation directory
-cd examples/activation
+# navigate to the relu directory
+cd examples/relu
 
 # then run the following command to build the kernel
 docker run --rm \
@@ -39,7 +39,7 @@ docker run --rm \
 ```
 
 This will build the kernel and save the output in the `build` directory in
-the activation folder.
+the relu folder.
 
 ## CLI Interface
 
@@ -55,10 +55,10 @@ The kernel builder includes a command-line interface for easier interaction. The
 ### Examples
 
 ```bash
-# Build the example activation kernel from the root of the repository
+# Build the example relu kernel from the root of the repository
 docker run --rm \
   -v $(pwd):/kernel-builder \
-  -w /kernel-builder/examples/activation \
+  -w /kernel-builder/examples/relu \
   ghcr.io/huggingface/kernel-builder:main \
   build
 
@@ -185,20 +185,20 @@ The whole goal of building these kernels is to allow researchers, developers, an
 To load a kernel locally, you should add the kernel build that is compatible with the Torch and CUDA versions in your environment to `PYTHONPATH`. For example:
 
 ```bash
-# PyTorch 2.6 and CUDA 12.6
-export PYTHONPATH="result/torch26-cxx11-cu126-x86_64-linux"
+# PyTorch 2.8 and CUDA 12.6
+export PYTHONPATH="result/torch28-cxx11-cu126-x86_64-linux"
 ```
 
 The kernel can then be imported as a Python module:
 
 ```python
 import torch
 
-import activation
+import relu
 
 x = torch.randn(10, 10)
 out = torch.empty_like(x)
-activation.silu_and_mul(out, x)
+relu.relu(x, out)
 
 print(out)
 ```

docs/nix.md

@@ -48,7 +48,7 @@ A kernel that has a `flake.nix` file can be built with the `build-and-copy`
 command. For example:
 
 ```bash
-cd examples/activation
+cd examples/relu
 nix run .#build-and-copy -L
 ```
 
@@ -94,7 +94,7 @@ with the kernel in Python's search path. This makes it more convenient to run
 tests:
 
 ```bash
-cd examples/activation
+cd examples/relu
 nix develop -L .#test
 python -m pytest tests
 ```
@@ -142,14 +142,3 @@ this check enabled, as it is one of the checks that validates that a kernel
 is compliant. This option is primarily intended for kernels with
 `triton.autotune` decorators, which can fail because there is no GPU available
 in the build sandbox.
-
-## Building a kernel without `flake.nix`
-
-If a kernels source directory does not have a `flake.nix` file, you can build the
-kernel using the `buildTorchExtensionBundle` function from the kernel builder
-itself:
-
-```bash
-cd examples/activation
-nix build --impure --expr 'with import ../..; lib.x86_64-linux.buildTorchExtensionBundle ./.' -L
-```

docs/writing-kernels.md

@@ -77,7 +77,7 @@ src = [
   "torch-ext/torch_binding.h"
 ]
 
-[kernel.activation]
+[kernel.relu]
 backend = "cuda"
 src = [
   "relu_kernel/relu.cu",