RoboFollow: Unveiling the Instruction Following Mirage in Embodied Agents
実装難易度
Easy
推論・学習コスト
High
想定用途
分類
概要
Modern embodied agents achieve impressive success rates, yet their actual instruction-following ability is far weaker than these numbers suggest. We trace this illusion to a structural property we term low scene entropy: when a visual scene admits only one valid task, language becomes redundant and a policy can score highly while barely using it. We introduce RoboFollow, a diagnostic benchmark…
何が新しいか
Modern embodied agents achieve impressive success rates, yet their actual instruction-following ability is far weaker than these numbers suggest. We trace this illusion to a structural property we…
何に使えるか
分類
実装情報
- GitHub URL
- あり
- Hugging Face URL
- あり
実装チェックリスト
実装または配布ページ
OKコードまたはモデル配布ページから検証を始められます。
一次情報リンク
OKHugging Face / GitHub
検証しやすさ
OK実装またはモデル配布ページから試せる可能性が高いです。
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まず自社データを、入力条件、目的変数、評価したい指標に分けて整理します。
- 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
Easy - 実装またはモデル配布ページから試せる可能性が高いです。
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: 推論だけで試せ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
UltraBench 2: Towards Robust Evaluation of Vision Foundation Models on Ultrasound
Benchmarking is an increasingly critical part of research in machine learning and the domains where it is appl
When Visual Quality Misleads: Intent Recognition under Rendered Avatar Distortions
Avatar-streaming systems are commonly evaluated with image and video quality assessment (IQA/VQA) metrics, imp
Geometric Mean Pooling for Equal-Weight Multiplicative Coarse-Graining
As an alternative to the additive and extremal biases of average and max pooling, we introduce Geometric Mean
How Many Labels Does Model Choice Need? Certificates and Budgets for Selective Prediction
Classifiers can make identical predictions yet require labels to compare their selective performance: confiden