開発者ツール
B
439 件中 181〜210 件目を表示
-
How we detect frontier model misalignmentOpenAI、フロンティアモデルの不整合を多層監視で検知OpenAIは、フロンティアモデルのミスアラインメントやエージェントの無許可の挙動を、多層的な監視で検出する取り組みを紹介した。インシデントが拡大する前に対応担当者が介入できるようにすることが狙い。
-
ChatGPTのPro・Businessプランで「API利用」可能に 月最大500ドル分ChatGPTのPro・Businessに月最大500ドル分のAPI利用枠を追加OpenAIはChatGPTのProおよびBusinessプランの特典として、同社APIを利用できる枠を追加すると発表した。月最大500ドル分（約7万9000円、1ドル158円換算）が付与され、ChatGPTの契約とAPI利用が結びつく形になる。
-
Does Federated Training Degrade Explainability? A Comparative Analysis of Grad-CAM Faithfulness in Centralized vs. Federated Deepfake Detection
-
Topology Survives Shuffling: Robust Long-Context Memory Access Prediction for Graph Analytics
-
CBE: A Complexity-adaptive Boundary-aware Ensemble Method for Imbalanced Data
-
Integrating Heterogeneous Echocardiography AI Outputs in Language Space: Learnable Late Fusion with a Supervised LLM
-
Sparse Fine-Tuning for Parameter-Efficient Adversarial Training
-
End-to-End Guarantees for Masked Diffusion Models under Conditional Mixing
-
Softmax Routing Dynamics in Mixture-of-Experts Models: Collapse and Recovery
-
ARBITRA: Robust Document-Level ATT&CK Extraction via Multi-Expert Argumentation
-
FDSML: Formula-Driven Supervised Metric Learning with Parameter-Aware Triplet Loss
-
Diagnosing the Reference Gap: What Visual Scaffolds Can and Cannot Fix in Multimodal Reasoning
-
LinQuant: Functional Invariances for Low-Precision Linear Recurrences
-
Well-Specified and Well-Tested? Assessing SWE-bench-Style Datasets with a Self-Evolving Agent
-
How Specific Do I Need to Be? Effects of User Query Underspecification on Deep Research Agents
-
Safe Multi-Objective Reinforcement Learning: Zero Constraint Violation and Primal-Averaging Convergence
-
Robustness of Chatbot LLM Judgments of Presidential Actions and Hypothetical Reforms
-
What Makes Medical Examination Items Difficult? Using Large Language Models to Explore Patterns Associated With Empirical Item Difficulty
-
Cord: Confidence-Anchored Order Residual Decoding for Masked Diffusion Language Models
-
More Is Better? Exploring Latent Correspondence in Federated Hypergraph Learning
-
Semantic Mismapping in Generalized Category Discovery: Evaluation and Mitigation
-
RetroChessBench: A Benchmark for Reconstructing and Proving Chess Histories
-
Does the Citation Matter? Quality, Grounding, and Physician-Rubric Accuracy of Google's AI Overviews for Medical Information
-
Class-aware Latent Attention Replay (CLAR): Class Incremental Hazard Diagnosis for Resilient AI in the Manufacturing Industrial Internet
-
BEVLight: Generalizable Traffic Signal Control from UAV Bird's-Eye Views
-
Learning to Bound: Monotonic, Calibrated Neural Upper Bounds for MPE Inference
-
There Is No Free Lunch in RLVR: Every pass@1 Gain Has a pass@k Price
-
Unlocking Any-Order Generation in Pretrained Autoregressive Image Models
-
Strongly Adaptive Online Learning Achieves Uniform Discounted Regret
-
Claude Code burns ~15–20% of your tokens on a 3-word status lineClaude Codeのステータス行が約15〜20%のトークンを消費との指摘HNで共有されたGist。Claude Codeが3語ほどのステータス行の生成に、トークン全体の約15〜20%を使っていると主張している。ページ題名によれば、agent_summaryを無効化するBunのプリロードをMITライセンスで公開している。数値の根拠や検証方法は未確認。