学習・ファインチューニング
A
118 件中 61〜90 件目を表示
-
Closed Gates Leak: Decomposing Adjacent-Layer Cancellation in Gated Language Models
-
When Does Closed-Loop Curriculum Adaptation Help GRPO? An Analysis on Countdown
-
Author-Informed Stance Detection: A Dataset and Findings
-
How Architecture, Typology, and Adaptation Strategies Shape ASR for Low-Resource African Languages: A Controlled Empirical Study
-
Beyond Passing Tests: A Systematic Review and Controlled Re-evaluation of Techniques for Improving the Quality of LLM-Generated Code
-
Interventional Training Reduces Causal Bypass in In-Context Introspection
-
Latent Decision Flow Rewards for Clinical Guideline Reasoning
-
Explore Where It Matters: Targeted Exploration for Efficient Rover Adaptation
-
オラクル、社員が業務アプリをAIで自動作成、Oracle Cloud上で共有できる「Oracle App Composer」発表。Oracle AI World 2026オラクル、AIで業務アプリを自動生成できる「Oracle App Composer」を発表オラクルは米ラスベガスで開催中の年次イベント「Oracle AI World 2026」で、社員が業務アプリをAIで自動生成し、Oracle Cloudにデプロイして利用できる「Oracle App Composer」を発表した。生成したアプリはOracle Cloud上で共有できる。なお詳細な仕様は本文未取得のため記載していない。
-
Curriculum-Based Deep Reinforcement Learning for Robust Autonomous Rocket Landing under Atmospheric Uncertainty
-
Co-Evolving Policy and Rubric via Capability Awareness for Open-Ended Generation
-
Persona Learning from Feedback for Personalized AI Assistants
-
PT$^3$: Prototype-Guided Self-Supervised Pretraining for Generalizable mmWave Human Point Cloud Representation
-
From Leniency to Suspicion: How an Open Post-Training Recipe Changes Whom LLM Fact-Checkers Believe
-
DES-JEPA: Learning a Shared Vocabulary for High-Dimensional Discrete Event Sequences
-
Predict the Reward, Not the World: World Reward Models for ML Agents
-
Discord、公開サーバの会話をGoogle検索などの対象に プライバシーへの懸念もDiscordが公開サーバの会話をWeb検索対象に、日本は初期設定オンでプライバシー懸念もDiscordは一部の公開サーバーの会話をGoogleなどのWeb検索の対象にすると発表した。11月中旬からAI要約を公開し、12月中旬には投稿本文も検索可能になる。日本では初期設定でオンとなるため、除外するには設定変更が必要。プライバシーやAI学習への利用を巡る懸念も広がっている。
-
Heads of Security: Learned Attention Gates Improve Prompt Injection Robustness
-
Can LLMs Uncover Hidden Physical Laws through Free Exploration?
-
MuonGS: Pretraining LLMs with Relaxed Weight-Norm Control
-
What Improves During Pretraining? Tracking Knowledge Acquisition with Point-in-Time Models
-
GAS-PFN: Graph-Aware Spatiotemporal Prior-Data Fitted Networks for Zero-Shot Traffic Forecasting
-
REGER: Modeling inter-item Relations with Semantic IDs for Generative Recommendation
-
When, Where, and What to Correct: Human-Guided Adaptive Residual Correction for Robot Policies
-
Oracle CloudがGoogle TPU導入を発表、2027年に提供を開始。Oracle AI World 2026Oracle CloudがGoogle TPUを導入、2027年からサービス提供開始オラクルはGoogleが開発するAI専用チップ「Google TPU」をOracle Cloud Infrastructureに導入し、2027年からGoogle TPUのサービスを提供開始すると発表した。発表はOracle AI World 2026の場で行われた。
-
A note on data-efficient finetuning of vision foundation models
-
Teach reality, not ethics: a drop-in continued-pretraining corpus installs coupled human-biosphere-AI welfare dispositions across model families
-
Follow the Gradient: Tracing Safety Degradation Back to Training Data
-
PerceptRL: Making Visual Perception Rewardable in MLLMs from Answers to Interactions
-
What Carries Forward? Pretraining Disclosure Across Language Model Releases