diff --git a/README.md b/README.md
index 796a8c9..397a62d 100644
--- a/README.md
+++ b/README.md
@@ -4,26 +4,21 @@
 
 ### MacOS
 
-1. Create virtual environment:
+#### 1. Setup Python:
 
 ```sh
 python3.8 -m venv venv8
 source venv8/bin/activate
-```
-
-2. Install dependencies:
-
-```sh
 pip3 install -r requirements-mac.txt
 ```
 
-3. Configure `plaidml` to use GPU:
+#### 2. Configure `plaidml` to use GPU:
 
 ```sh
 plaidml-setup
 ```
 
-4. Configure environment variables in the run script:
+#### 3. Configure environment variables in the run script:
 
 ```sh
 export KERAS_BACKEND="plaidml.keras.backend"
@@ -32,7 +27,7 @@ export tg_token="Your telegram token here"
 
 ### Windows (CUDA)
 
-1. Setup Python
+#### 1. Setup Python
 
 ```powershell
 python3.9 -m venv venv
@@ -40,7 +35,7 @@ python3.9 -m venv venv
 pip install -r requirements-win-cuda.txt
 ```
 
-2. Install CUDA
+#### 2. Install CUDA
 
 * Install NVIDIA Drivers: https://www.nvidia.com/drivers
 * Install CUDA **11.2** (for TensorFlow 2.7.0): https://developer.nvidia.com/cuda-toolkit-archive
@@ -48,7 +43,7 @@ pip install -r requirements-win-cuda.txt
   * Copy folders in `cudnn-11.2-windows-x64-v8.1.1.33\cuda` to `C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.2`
 * Restart IntelliJ IDEA
 
-3. Check Device List
+#### 3. Check Device List
 
 ```shell
 python -c "from tensorflow.python.client import device_lib; print(device_lib.list_local_devices())"