docs: add README for PyTorch examples with Kubeflow Trainer SDK

Author: Snehadas2005

examples/pytorch/README.md

@@ -0,0 +1,33 @@
+# PyTorch Examples
+
+This directory contains examples for training PyTorch models using the Kubeflow Trainer SDK.
+
+### Examples
+
+| Use Case | Script | Notebook |
+| :--- | :--- | :--- |
+| **Image Classification** | [train_mnist.py](image-classification/train_mnist.py) | [mnist.ipynb](image-classification/mnist.ipynb) |
+
+### Prerequisites
+
+To run these examples, install the Kubeflow SDK:
+```bash
+pip install -U kubeflow
+```
+
+### How to Run
+
+These standalone scripts are designed for automated workflows and production training. They automatically handle distributed setup and dependency installation on the cluster.
+
+**Submit an MNIST training job:**
+```bash
+python image-classification/train_mnist.py --nodes 1
+```
+
+**Verify locally (no Kubernetes needed):**
+You can verify the training logic on your local machine using the `--test` flag:
+```bash
+python image-classification/train_mnist.py --test
+```
+
+For interactive experimentation, you can also use the corresponding Jupyter notebooks in each subdirectory.