diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml
index 0435884..6a3921d 100644
--- a/.pre-commit-config.yaml
+++ b/.pre-commit-config.yaml
@@ -55,7 +55,7 @@ repos:
       - id: python-use-type-annotations
       - id: text-unicode-replacement-char
   - repo: https://github.com/codespell-project/codespell
-    rev: v2.3.0
+    rev: v2.4.0
     hooks:
       - id: codespell
         additional_dependencies: ["tomli"]
diff --git a/README.md b/README.md
index 7fc1605..fcace5e 100644
--- a/README.md
+++ b/README.md
@@ -49,6 +49,21 @@ pip install uv
 
 ## Running the Benchmark
 
+To reproduce the benchmarks, you'll need the ImageNet validation dataset:
+
+Download the validation set (50,000 images, ~6.3GB):
+
+```bash
+wget https://image-net.org/data/ILSVRC/2012/ILSVRC2012_img_val.tar
+```
+
+Create a directory and extract the images:
+
+```bash
+mkdir -p imagenet/val
+tar -xf ILSVRC2012_img_val.tar -C imagenet/val
+```
+
 The benchmark script creates separate virtual environments for each library and
 runs tests independently: