{
  "id": 674730,
  "title": "Final Online Round Evaluation — Submit Your Final Notebooks",
  "url": "/competitions/dl-sprint-4-0-bengali-long-form-speech-recognition/discussion/674730",
  "author_name": "Istiak Ahmmed Rifti",
  "post_date": "2026-02-21T20:07:56.682000",
  "votes": 0,
  "comment_count": 0,
  "views": 0,
  "content": "<p>As we move into the <strong>final evaluation phase</strong>, every team should submit their <strong>final notebooks</strong>.</p>\n<h3>What to submit</h3>\n<ul>\n<li><strong>All training-related notebooks</strong> (e.g., training/fine-tuning, preprocessing, EDA/feature engineering and any other implementation/changes).</li>\n<li><strong>Only ONE final inference notebook</strong> (this will be used to generate predictions and run on the hidden test set). Please clearly mention which notebook is the <strong>final inference notebook</strong>.</li>\n</ul>\n<p>We will <strong>re-run all submitted notebooks</strong> and <strong>re-evaluate</strong> the results. The scores from this re-run will be considered the <strong>final Public and Private leaderboard scores</strong> (overriding what you currently see on the leaderboard).</p>\n<p>In addition, we will run the <strong>final inference notebook</strong> on a <strong>hidden test set</strong> and record the score.<br>\nBased on the <strong>re-run Public score + re-run Private score + hidden set score</strong>, we will <strong>shortlist the top teams</strong> for the next stage.</p>\n<p>Thank you and best of luck!</p>",
  "messages": [
    {
      "id": 3409003,
      "postDate": "2026-02-21T20:07:56.683Z",
      "content": "<p>As we move into the <strong>final evaluation phase</strong>, every team should submit their <strong>final notebooks</strong>.</p>\n<h3>What to submit</h3>\n<ul>\n<li><strong>All training-related notebooks</strong> (e.g., training/fine-tuning, preprocessing, EDA/feature engineering and any other implementation/changes).</li>\n<li><strong>Only ONE final inference notebook</strong> (this will be used to generate predictions and run on the hidden test set). Please clearly mention which notebook is the <strong>final inference notebook</strong>.</li>\n</ul>\n<p>We will <strong>re-run all submitted notebooks</strong> and <strong>re-evaluate</strong> the results. The scores from this re-run will be considered the <strong>final Public and Private leaderboard scores</strong> (overriding what you currently see on the leaderboard).</p>\n<p>In addition, we will run the <strong>final inference notebook</strong> on a <strong>hidden test set</strong> and record the score.<br>\nBased on the <strong>re-run Public score + re-run Private score + hidden set score</strong>, we will <strong>shortlist the top teams</strong> for the next stage.</p>\n<p>Thank you and best of luck!</p>",
      "rawMarkdown": "As we move into the **final evaluation phase**, every team should submit their **final notebooks**.\n\n### What to submit\n- **All training-related notebooks** (e.g., training/fine-tuning, preprocessing, EDA/feature engineering and any other implementation/changes).\n- **Only ONE final inference notebook** (this will be used to generate predictions and run on the hidden test set). Please clearly mention which notebook is the **final inference notebook**.\n\nWe will **re-run all submitted notebooks** and **re-evaluate** the results. The scores from this re-run will be considered the **final Public and Private leaderboard scores** (overriding what you currently see on the leaderboard).\n\nIn addition, we will run the **final inference notebook** on a **hidden test set** and record the score.  \nBased on the **re-run Public score + re-run Private score + hidden set score**, we will **shortlist the top teams** for the next stage.\n\nThank you and best of luck!"
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "3409003": "As we move into the **final evaluation phase**, every team should submit their **final notebooks**.\n\n### What to submit\n- **All training-related notebooks** (e.g., training/fine-tuning, preprocessing, EDA/feature engineering and any other implementation/changes).\n- **Only ONE final inference notebook** (this will be used to generate predictions and run on the hidden test set). Please clearly mention which notebook is the **final inference notebook**.\n\nWe will **re-run all submitted notebooks** and **re-evaluate** the results. The scores from this re-run will be considered the **final Public and Private leaderboard scores** (overriding what you currently see on the leaderboard).\n\nIn addition, we will run the **final inference notebook** on a **hidden test set** and record the score.  \nBased on the **re-run Public score + re-run Private score + hidden set score**, we will **shortlist the top teams** for the next stage.\n\nThank you and best of luck!"
  }
}