Typo

Author: awf

docs/source/04-benchmark.ipynb

@@ -2,18 +2,9 @@
  "cells": [
   {
    "cell_type": "code",
-   "execution_count": 12,
+   "execution_count": 1,
    "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "The autoreload extension is already loaded. To reload it, use:\n",
-      "  %reload_ext autoreload\n"
-     ]
-    }
-   ],
+   "outputs": [],
    "source": [
     "# Copyright (c) 2024 Graphcore Ltd. All rights reserved.\n",
     "\n",
@@ -39,24 +30,31 @@
     "\n",
     "The `gfloat` library is designed for readability over performance, and the reference code for computations is the (slow) scalar code e.g. `round_float`.\n",
     "\n",
-    "There are vectorized implementations (e.g. `round_ndarray`, and when combined with JAX, these can go reasonably fast).\n",
+    "There are vectorized implementations (e.g. `round_ndarray`), and when combined with JAX, these can go reasonably fast.\n",
     "\n",
     "Let's see how long it takes to encoed some values to FP8..."
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 2,
    "metadata": {},
    "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "An NVIDIA GPU may be present on this machine, but a CUDA-enabled jaxlib is not installed. Falling back to cpu.\n"
+     ]
+    },
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "GFloat scalar                  :  6996.75 nsec (50 runs at size 10000)\n",
-      "GFloat vectorized, numpy arrays:    75.04 nsec (50 runs at size 1000000)\n",
-      "GFloat vectorized, JAX JIT     :     3.18 nsec (1000 runs at size 1000000)\n",
-      "ML_dtypes                      :     3.13 nsec (1000 runs at size 1000000)\n"
+      "GFloat scalar                  :  6666.04 nsec (50 runs at size 10000)\n",
+      "GFloat vectorized, numpy arrays:    57.84 nsec (50 runs at size 1000000)\n",
+      "GFloat vectorized, JAX JIT     :     3.17 nsec (1000 runs at size 1000000)\n",
+      "ML_dtypes                      :     2.92 nsec (1000 runs at size 1000000)\n"
      ]
     }
    ],