Upload hf_readme.ipynb

hf_readme.ipynb

@@ -0,0 +1,228 @@
+{
+  "nbformat": 4,
+  "nbformat_minor": 0,
+  "metadata": {
+    "colab": {
+      "provenance": []
+    },
+    "kernelspec": {
+      "name": "python3",
+      "display_name": "Python 3"
+    },
+    "language_info": {
+      "name": "python"
+    }
+  },
+  "cells": [
+    {
+      "cell_type": "code",
+      "execution_count": 15,
+      "metadata": {
+        "id": "mSG4QdpNvio4"
+      },
+      "outputs": [],
+      "source": []
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "!pip install kenlm"
+      ],
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "f3jTUbzZv442",
+        "outputId": "76eb7e32-227b-4cef-a20f-38ae936c4d8a"
+      },
+      "execution_count": 16,
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stdout",
+          "text": [
+            "Requirement already satisfied: kenlm in /usr/local/lib/python3.12/dist-packages (0.3.0)\n"
+          ]
+        }
+      ]
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "from huggingface_hub import hf_hub_download\n",
+        "import kenlm\n",
+        "\n",
+        "# Step 1: Download the .arpa file from the dataset repository\n",
+        "print(\"Downloading KenLM ARPA file from Hugging Face Dataset...\")\n",
+        "model_path = hf_hub_download(\n",
+        "    repo_id=\"BeitTigreAI/tigre-data-kenLM\",\n",
+        "    filename=\"tigre-data-kenLM.arpa\",\n",
+        "    repo_type=\"dataset\",  # 🔑 This is critical! It's a dataset repo, not a model\n",
+        "    local_dir=\"./kenlm_model\"\n",
+        ")\n",
+        "\n"
+      ],
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "Q8KTpqAqvj4d",
+        "outputId": "6fface0f-230d-49a6-ea02-02ab06850f46"
+      },
+      "execution_count": 17,
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stdout",
+          "text": [
+            "Downloading KenLM ARPA file from Hugging Face Dataset...\n"
+          ]
+        }
+      ]
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "import kenlm\n",
+        "\n",
+        "# Load the language model\n",
+        "model = kenlm.Model(model_path)\n",
+        "\n",
+        "# Example sentences\n",
+        "sentences = [\n",
+        "    \"አነ ምሩር ህሌኮ ምን ኪደትለ አፌተ\",\n",
+        "    \"እብ ረቢኩም ምን ተአምኖ፡ ለወለት ልብዬ ወዕንቼተ\",\n",
+        "    \"ህተ ትብል ሑዬቱ ወአነ እብል ሕቼተ\"\n",
+        "]\n",
+        "\n",
+        "def test_sentence(sentence):\n",
+        "    log_prob = model.score(sentence)\n",
+        "    perplexity = model.perplexity(sentence)\n",
+        "    print(f\"Sentence: {sentence}\")\n",
+        "    print(f\"Log10 Probability: {log_prob:.4f}\")\n",
+        "    print(f\"Perplexity: {perplexity:.4f}\")\n",
+        "    print(\"-\" * 50)\n",
+        "\n",
+        "for sent in sentences:\n",
+        "    test_sentence(sent)"
+      ],
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "UHem-Q4Kvslb",
+        "outputId": "d9a947a0-01ae-402f-e592-0a9c6951d962"
+      },
+      "execution_count": 18,
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stdout",
+          "text": [
+            "Sentence: አነ ምሩር ህሌኮ ምን ኪደትለ አፌተ\n",
+            "Log10 Probability: -27.5345\n",
+            "Perplexity: 8580.2515\n",
+            "--------------------------------------------------\n",
+            "Sentence: እብ ረቢኩም ምን ተአምኖ፡ ለወለት ልብዬ ወዕንቼተ\n",
+            "Log10 Probability: -34.3204\n",
+            "Perplexity: 19500.8208\n",
+            "--------------------------------------------------\n",
+            "Sentence: ህተ ትብል ሑዬቱ ወአነ እብል ሕቼተ\n",
+            "Log10 Probability: -28.4631\n",
+            "Perplexity: 11645.5115\n",
+            "--------------------------------------------------\n"
+          ]
+        }
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "source": [
+        "<!DOCTYPE html>\n",
+        "<html lang=\"en\">\n",
+        "<head>\n",
+        "  <meta charset=\"UTF-8\" />\n",
+        "  <meta name=\"viewport\" content=\"width=device-width, initial-scale=1.0\"/>\n",
+        "  <title>Tigre LM Output Explanation</title>\n",
+        "  <style>\n",
+        "    body {\n",
+        "      font-family: Arial, sans-serif;\n",
+        "      line-height: 1.6;\n",
+        "      color: #333;\n",
+        "      max-width: 600px;\n",
+        "      margin: 2rem auto;\n",
+        "      padding: 1rem;\n",
+        "    }\n",
+        "    h2 {\n",
+        "      color: #2c3e50;\n",
+        "      border-bottom: 2px solid #3498db;\n",
+        "      padding-bottom: 0.3em;\n",
+        "    }\n",
+        "    table {\n",
+        "      width: 100%;\n",
+        "      border-collapse: collapse;\n",
+        "      margin: 1.2em 0;\n",
+        "    }\n",
+        "    th, td {\n",
+        "      padding: 0.8em;\n",
+        "      text-align: left;\n",
+        "      border-bottom: 1px solid #ddd;\n",
+        "    }\n",
+        "    th {\n",
+        "      background-color: #f2f2f2;\n",
+        "      font-weight: bold;\n",
+        "    }\n",
+        "    ul {\n",
+        "      margin: 0.5em 0;\n",
+        "      padding-left: 1.5em;\n",
+        "    }\n",
+        "    .checkmark {\n",
+        "      color: #27ae60;\n",
+        "      font-weight: bold;\n",
+        "    }\n",
+        "  </style>\n",
+        "</head>\n",
+        "<body>\n",
+        "\n",
+        "  <h2>Understanding the Output</h2>\n",
+        "\n",
+        "  <table>\n",
+        "    <tr>\n",
+        "      <th>Score</th>\n",
+        "      <th>Meaning</th>\n",
+        "      <th>Ideal</th>\n",
+        "    </tr>\n",
+        "    <tr>\n",
+        "      <td><strong>Log10 Probability</strong></td>\n",
+        "      <td>Log of sentence probability. Higher (less negative) = more fluent.</td>\n",
+        "      <td>Closer to 0</td>\n",
+        "    </tr>\n",
+        "    <tr>\n",
+        "      <td><strong>Perplexity</strong></td>\n",
+        "      <td>Measures surprise. Lower = more natural.</td>\n",
+        "      <td>As low as possible</td>\n",
+        "    </tr>\n",
+        "    <tr>\n",
+        "      <td><strong>OOV</strong><br><em>(in word scores)</em></td>\n",
+        "      <td><code>True</code> = word not in vocabulary. Affects confidence.<br>\n",
+        "          <code>False</code> = word is known.</td>\n",
+        "      <td><code>False</code></td>\n",
+        "    </tr>\n",
+        "  </table>\n",
+        "\n",
+        "  <p><span class=\"checkmark\">✅ Use this to:</span></p>\n",
+        "  <ul>\n",
+        "    <li>Compare fluency of different sentences</li>\n",
+        "    <li>Rescore ASR outputs</li>\n",
+        "    <li>Detect unnatural or grammatically odd phrases</li>\n",
+        "  </ul>\n",
+        "\n",
+        "</body>\n",
+        "</html>"
+      ],
+      "metadata": {
+        "id": "AF6K4_3R0wlM"
+      }
+    }
+  ]
+}