improve formatting

ch05/07_gpt_to_llama/converting-gpt-to-llama2.ipynb

@@ -39,10 +39,10 @@
     "id": "aFmxTQbwCUMl"
    },
    "source": [
-    "- In this notebook, we convert the original GPT and GPT-2 architecture into a Llama 2 model step by step\n",
+    "- In this notebook, we convert the original GPT architecture into a Llama 2 model step by step (note the GPT and GPT-2 share the same architecture)\n",
     "- Why not Llama 1 or Llama 3?\n",
-    " - The Llama 1 architecture is similar to Llama 2, except that Llama 2 has a larger context window (which is nice); the Llama 1 weights are not readily available and have more usage restrictions, so it makes more sense to focus on Llama 2\n",
-    " - Regarding Llama 3, I will share a separate notebook to convert Llama 2 to Llama 3 (there are only a few small additional changes)\n",
+    "   - The Llama 1 architecture is similar to Llama 2, except that Llama 2 has a larger context window (which is nice); the Llama 1 weights are not readily available and have more usage restrictions, so it makes more sense to focus on Llama 2\n",
+    "   - Regarding Llama 3, I will share a separate notebook to convert Llama 2 to Llama 3 (there are only a few small additional changes)\n",
     "- The explanations are purposefully kept minimal in this notebook not to bloat it unnecessarily and focus on the main code\n",
     "- For more information, please see the Llama 2 paper: [Llama 2: Open Foundation and Fine-Tuned Chat Models (2023)](https://arxiv.org/abs/2307.09288)"
    ]
@@ -143,7 +143,7 @@
     "- LayerNorm normalizes inputs using mean and variance, while RMSNorm uses only the root mean square, which improves computational efficiency\n",
     "- The RMSNorm operation is as follows, where $x$ is the input $\\gamma$ is a trainable parameter (vector), and $\\epsilon$ is a small constant to avoid zero-division errors:\n",
     "\n",
-    "$$y = \\frac{x}{\\sqrt{\\text{RMS}[x^2]} + \\epsilon} * \\gamma$$\n",
+    "$$y_i = \\frac{x_i}{\\text{RMS}(x)} \\gamma_i, \\quad \\text{where} \\quad \\text{RMS}(x) = \\sqrt{\\epsilon + \\frac{1}{n} \\sum x_i^2}$$\n",
     "\n",
     "- For more details, please see the paper [Root Mean Square Layer Normalization (2019)](https://arxiv.org/abs/1910.07467)"
    ]
@@ -1565,7 +1565,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.10.11"
+   "version": "3.11.4"
   }
  },
  "nbformat": 4,