diff --git a/RELEASE.md b/RELEASE.md
index b248bae9a68..02550040c47 100644
--- a/RELEASE.md
+++ b/RELEASE.md
@@ -177,7 +177,7 @@ stjohnso98, <NAME>, <HERE>, <USING>, <GITHUB>, <HANDLE>
     * Update `tf.saved_model.SaveOptions` with [`experimental_io_device`](https://www.tensorflow.org/versions/r2.3/api_docs/python/tf/saved_model/SaveOptions?hl=en) as arg with default value `None` to choose the I/O device for saving models and weights.
     * Mutable tables now restore checkpointed values when loaded from SavedModel.
   * GPU
-    * No longer includes PTX kernels for GPU except for sm_70 to reduce binary size. On systems with NVIDIA® Ampere GPUs (CUDA architecture 8.0) or newer, kernels are JIT-compiled from PTX and TensorFlow can take over 30 minutes to start up. This overhead can be limited to the first start up by increasing the default JIT cache size with: `export CUDA_CACHE_MAXSIZE=2147483648`.:
+    * TF 2.3 includes PTX kernels only for [compute capability](https://developer.nvidia.com/cuda-gpus) 7.0 to reduce the TF pip binary size.  Earlier releases included PTX for a variety of older compute capabilities.
   * Others
     * Retain parent namescope for ops added inside `tf.while_loop`/`tf.cond`/`tf.switch_case`.
     * Update `tf.vectorized_map` to support vectorizing `tf.while_loop` and TensorList operations.