diff --git a/burn-tensor/README.md b/burn-tensor/README.md
index 37501a301a..f7f6855419 100644
--- a/burn-tensor/README.md
+++ b/burn-tensor/README.md
@@ -50,7 +50,7 @@ Therefore, creating the tape only requires a simple and efficient graph traversa
 
 ## Cuda
 
-To run with CUDA set `TORCH_CUDA_VERSION=cu113`.
+To run with CUDA set `TORCH_CUDA_VERSION=cu121`.
 
 ## Notes
 
diff --git a/examples/mnist/README.md b/examples/mnist/README.md
index 4040dc56a6..4aa61016d4 100644
--- a/examples/mnist/README.md
+++ b/examples/mnist/README.md
@@ -17,7 +17,7 @@ cargo run --example mnist --release --features ndarray                # CPU NdAr
 cargo run --example mnist --release --features ndarray-blas-openblas  # CPU NdArray Backend - f32 - blas with openblas
 cargo run --example mnist --release --features ndarray-blas-netlib    # CPU NdArray Backend - f32 - blas with netlib
 echo "Using tch backend"
-export TORCH_CUDA_VERSION=cu113                                       # Set the cuda version
+export TORCH_CUDA_VERSION=cu121                                       # Set the cuda version
 cargo run --example mnist --release --features tch-gpu                # GPU Tch Backend - f32
 cargo run --example mnist --release --features tch-cpu                # CPU Tch Backend - f32
 echo "Using wgpu backend"
diff --git a/examples/simple-regression/README.md b/examples/simple-regression/README.md
index 4c385fea70..4f4f445c01 100644
--- a/examples/simple-regression/README.md
+++ b/examples/simple-regression/README.md
@@ -23,7 +23,7 @@ cargo run --example regression --release --features ndarray                # CPU
 cargo run --example regression --release --features ndarray-blas-openblas  # CPU NdArray Backend - f32 - blas with openblas
 cargo run --example regression --release --features ndarray-blas-netlib    # CPU NdArray Backend - f32 - blas with netlib
 echo "Using tch backend"
-export TORCH_CUDA_VERSION=cu113                                       # Set the cuda version
+export TORCH_CUDA_VERSION=cu121                                            # Set the cuda version
 cargo run --example regression --release --features tch-gpu                # GPU Tch Backend - f32
 cargo run --example regression --release --features tch-cpu                # CPU Tch Backend - f32
 echo "Using wgpu backend"
diff --git a/examples/text-classification/README.md b/examples/text-classification/README.md
index a38e78f0b9..cefd5e3784 100644
--- a/examples/text-classification/README.md
+++ b/examples/text-classification/README.md
@@ -29,7 +29,7 @@ cd burn
 # Use the --release flag to really speed up training.
 # Use the f16 feature if your CUDA device supports FP16 (half precision) operations. May not work well on every device.
 
-export TORCH_CUDA_VERSION=cu117  # Set the cuda version (CUDA users)
+export TORCH_CUDA_VERSION=cu121  # Set the cuda version (CUDA users)
 
 # AG News
 cargo run --example ag-news-train --release --features tch-gpu   # Train on the ag news dataset
diff --git a/examples/text-generation/README.md b/examples/text-generation/README.md
index a8953244ac..0131c0c3e2 100644
--- a/examples/text-generation/README.md
+++ b/examples/text-generation/README.md
@@ -14,7 +14,7 @@ git clone https://github.com/tracel-ai/burn.git
 cd burn
 
 # Use the --release flag to really speed up training.
-export TORCH_CUDA_VERSION=cu113
+export TORCH_CUDA_VERSION=cu121
 cargo run --example text-generation --release
 ```