Hubs or Fringes: Pretraining Data Selection via Web Graph Centrality
Vedant Badoni, Danqi Chen, Xinyi Wang
実装難易度
Hard
推論・学習コスト
High
想定用途
分類
概要
The performance of modern language models depends critically on pretraining data composition. Yet existing data selection methods rely on auxiliary classifiers for document scoring or mixture optimization, adding computational overhead and dependence on labeled data. We propose WebGraphMix, a lightweight data selection framework that computes structural centrality scores over the Common Crawl…
何が新しいか
The performance of modern language models depends critically on pretraining data composition. Yet existing data selection methods rely on auxiliary classifiers for document scoring or mixture…
何に使えるか
分類
実装情報
- Paper URL
- あり
実装チェックリスト
実装または配布ページ
要確認Paper onlyの可能性があるため再実装前提で確認してください。
一次情報リンク
OKPaper
検証しやすさ
要確認公式実装が見つからないため、論文から再実装する前提です。
計算資源
要確認学習や高解像度推論ではGPUメモリと実行時間に注意が必要です。
ライセンス
未取得配布元のLICENSE、モデルカード、Paperの利用条件を確認してください。
商用利用
未取得研究利用限定、データセット由来制限、API規約の有無を確認してください。
自社データで試すなら
製造業・材料開発のExcel/CSVデータに落とし込むための最初の手順です。
- 1まずExcel/CSVの実験条件、組成、物性値を1行1実験の表に整理します。
- 2LightGBMやRandom Forestなどのベースラインを先に作り、この手法と比較します。
- 3評価指標はR2/RMSE、AUC、異常検知の再現率、実験回数削減率など、現場の意思決定に近いものを選びます。
- 4SHAPや特徴量重要度で、効いている因子が物理・化学・工程知識と矛盾しないか確認します。
実装難易度
Hard - 公式実装が見つからないため、論文から再実装する前提です。
必要リソース
- GPU目安: High
- データセット: 論文・リポジトリ側の指定を確認してください。
- 学習要否: 再学習や評価環境の準備が必要になる可能性があります。
- 学習や高解像度推論ではGPUメモリと実行時間に注意が必要です。
実務で使う場合の注意点
- ライセンスと商用利用条件は、Paper / GitHub / Hugging Face の配布元で確認してください。
- 精度、再現性、計算コストはデータセットや評価条件に依存します。
- 個人情報や機密データを扱う場合は、入力データの保存先と外部API利用条件を確認してください。
関連記事
RCAP: Robust, Class-Aware, Probabilistic Dynamic Dataset Pruning
ダイナミックデータ削減技術を適用する際に起きる、バランスの取れたデータにおける損失が大きくなる問題を解決するために、新しいクラス認識付きの確率的データ削減手法を提案した。
When is Your LLM Steerable?
この研究では、LLMの制御を目指す研究を行います。
Accurate and Resource-Efficient Federated Continual Learning
Federated continual learning (FCL) must learn from distributed task streams under limited resources, such as c
Towards Fully Automated Exam Grading: Fairness-Aware Recognition of Handwritten Answers with Foundation Models
Correcting handwritten exams by hand is time-consuming and error-prone, particularly for large cohorts, while