{
 "log": "LOG480",
 "article_full": "/files/LOG480/LOG480_full.md",
 "article_full_type": "text/markdown",
 "zip": "/files/LOG480/LOG480_files.zip",
 "files": [
  {
   "file": "01_sec07.txt",
   "url": "/files/LOG480/01_sec07.txt",
   "chars": 4631,
   "lines": 54,
   "lang": "text",
   "section": "依頼133：base から学習する deep1 を足して本番を始めた",
   "heading": "依頼133 の報告（r2r2r-scripts/claude_code_reports/20261109_133.txt）",
   "subheading": ""
  },
  {
   "file": "02_commit.txt",
   "url": "/files/LOG480/02_commit.txt",
   "chars": 57904,
   "lines": 809,
   "lang": "diff",
   "section": "依頼133：base から学習する deep1 を足して本番を始めた",
   "heading": "依頼133 の commit（8db7028）の差分の全量",
   "subheading": ""
  },
  {
   "file": "03_sec08.txt",
   "url": "/files/LOG480/03_sec08.txt",
   "chars": 53789,
   "lines": 1176,
   "lang": "text",
   "section": "学習の記録 ― Runpod の A40 で3,000ステップ（KIOXIA の lab/Log27/train/）",
   "heading": "学習のターミナルの記録（lab_train_20261109_153622.txt）",
   "subheading": ""
  },
  {
   "file": "04_sec08.json",
   "url": "/files/LOG480/04_sec08.json",
   "chars": 4275,
   "lines": 132,
   "lang": "json",
   "section": "学習の記録 ― Runpod の A40 で3,000ステップ（KIOXIA の lab/Log27/train/）",
   "heading": "由来の記録（train_provenance.json）",
   "subheading": ""
  },
  {
   "file": "05_sec08.txt",
   "url": "/files/LOG480/05_sec08.txt",
   "chars": 30728,
   "lines": 577,
   "lang": "text",
   "section": "学習の記録 ― Runpod の A40 で3,000ステップ（KIOXIA の lab/Log27/train/）",
   "heading": "LeRobot の形への変換と正規化の値の測り直しのログ（prep_deep.log）",
   "subheading": ""
  },
  {
   "file": "06_sec08.txt",
   "url": "/files/LOG480/06_sec08.txt",
   "chars": 239705,
   "lines": 1920,
   "lang": "text",
   "section": "学習の記録 ― Runpod の A40 で3,000ステップ（KIOXIA の lab/Log27/train/）",
   "heading": "学習のログ（train_deep_even10_v1.log）",
   "subheading": ""
  },
  {
   "file": "07_sec08.txt",
   "url": "/files/LOG480/07_sec08.txt",
   "chars": 57528,
   "lines": 729,
   "lang": "text",
   "section": "学習の記録 ― Runpod の A40 で3,000ステップ（KIOXIA の lab/Log27/train/）",
   "heading": "学習中の GPU の記録・30秒ごと（gpu_train.csv）",
   "subheading": ""
  },
  {
   "file": "08_sec08.txt",
   "url": "/files/LOG480/08_sec08.txt",
   "chars": 16154,
   "lines": 240,
   "lang": "text",
   "section": "学習の記録 ― Runpod の A40 で3,000ステップ（KIOXIA の lab/Log27/train/）",
   "heading": "Pod 側のチェックポイントの一覧（pod_checkpoints.tsv）",
   "subheading": ""
  },
  {
   "file": "09_sec09.txt",
   "url": "/files/LOG480/09_sec09.txt",
   "chars": 4398,
   "lines": 69,
   "lang": "text",
   "section": "依頼135：安全範囲を50本から作り、次の学習から学習率をステップ数に合わせた",
   "heading": "依頼135 の報告（r2r2r-scripts/claude_code_reports/20261109_135.txt）",
   "subheading": ""
  },
  {
   "file": "10_commit.txt",
   "url": "/files/LOG480/10_commit.txt",
   "chars": 65348,
   "lines": 795,
   "lang": "diff",
   "section": "依頼135：安全範囲を50本から作り、次の学習から学習率をステップ数に合わせた",
   "heading": "依頼135 の commit（6715f5e）の差分の全量",
   "subheading": ""
  },
  {
   "file": "11_commit.txt",
   "url": "/files/LOG480/11_commit.txt",
   "chars": 15855,
   "lines": 228,
   "lang": "diff",
   "section": "依頼137：写り方の確かめの基準画像を trim2 にそろえた",
   "heading": "依頼137 の commit（3310ac0）の差分の全量",
   "subheading": ""
  },
  {
   "file": "12_sec11.py.txt",
   "url": "/files/LOG480/12_sec11.py.txt",
   "chars": 5597,
   "lines": 118,
   "lang": "python",
   "section": "依頼137 のあとの、base から学習する2つのファイルの全体",
   "heading": "学習の設定を openpi に足すコード（r2r2r-scripts/pi05/openpi_addon/add_deep_config.py）",
   "subheading": ""
  },
  {
   "file": "13_sec11.sh.txt",
   "url": "/files/LOG480/13_sec11.sh.txt",
   "chars": 3633,
   "lines": 74,
   "lang": "bash",
   "section": "依頼137 のあとの、base から学習する2つのファイルの全体",
   "heading": "Pod でデータを変換し正規化の値を測り直すスクリプト（r2r2r-scripts/pi05/openpi_addon/pod_prep_deep.sh）",
   "subheading": ""
  }
 ],
 "article_full_markdown": "/files/LOG480/LOG480_full.md"
}