diff a/hat/docs/Examples/examples.md b/hat/docs/Examples/examples.md
--- /dev/null
+++ b/hat/docs/Examples/examples.md
@@ -0,0 +1,77 @@
+# Running HAT Examples
+[Back to Index ../](../index.md)
+
+The [examples-package]() in HAT contains a list of examples varying from matrix operations, DFT, Flash-Attention, nbody, shaders, and image detection.
+
+To run an example:
+
+```bash
+java @.ffi-<backend>-example <example>
+```
+
+For instance, to run `flash-attention` with the OpenCL backend:
+
+```bash
+java @.ffi-opencl-example flashattention.Main
+```
+For the CUDA backend:
+```bash
+java @.ffi-cuda-example  flashattention.Main
+```
+
+## Options for Examples
+
+Some of the examples accept command line options to specify input size, kernel version, etc.
+
+For example, `flashattention`:
+
+```bash
+java @.ffi-opencl-example flashattention.Main --size=2048 --iterations=10 --verbose
+```
+
+For flash attention You can see the full list of options by using `--help`:
+
+```bash
+    --size=<size>                   Specify an input size
+    --iterations=<numIterations>    Specify the number of iterations to perform
+    --skip-sequential               Flag to bypass the sequential execution in Java
+    --check                         Flag to check the results. This implies the Java sequential version runs.
+    --verbose                       Flag to print information between runs (e.g., total time).
+    --help                          Print this help.
+```
+
+### Obtaining information about the accelerator used to launch the application
+
+You can use the variable `INFO` to indicate the HAT runtime to dump the device name and device version used to launch the generated GPU kernel:
+
+
+```bash
+$ HAT=INFO java @.ffi-cuda-example matmul.Main
+
+[INFO] Input Size     : 1024x1024
+[INFO] Check Result:  : false
+[INFO] Num Iterations : 100
+[INFO] NDRangeConfiguration: 2DTILING
+
+[INFO] Using NVIDIA GPU: NVIDIA GeForce RTX 5060   << an NVIDIA 5060 was used
+[INFO] Dispatching the CUDA kernel
+        \_ BlocksPerGrid   = [64,64,1]    << Num blocks dispatched
+        \_ ThreadsPerBlock = [16,16,1]    << threads-per-block dispatched
+```
+
+## Running Headless Versions
+
+Most GUI applications contains a `headless` version, which can be able by setting the following
+
+```bash
+java @.ffi-opencl-example -Dheadless=true  mandel.Main
+```
+
+Alternatively you can usually pass an arg to the app itself
+
+```bash
+java @.ffi-opencl-example  mandel.Main --headless
+```
+
+This sets `-Dheadless=true` and passes '--headless' to the example.
+