【AI最新動向 2026年2月15日】論文5件・GitHub5件

2026年2月15日2026年3月27日

📝 この記事のポイント

🚀 AI技術の最新動向 – 2026年2月15日世界中から収集したAI・機械学習の最新情報をお届けします 📑 目次 💻 注目のGitHubプロジェクト milanm/AutoGrad-Engine Kuberwastaken/picogpt ScottT2-spec/malaria-cell-detection NYX-VORAX/lightning-image-scraper moketchups/permanently-jailbroken 📌 関連記事もチェック 📚 最新研究論文 1. Scaling Verification Can Be More Effective than Scaling Policy Learning for Vision-Language-Action Alignment 著者: Jacky Kwok, Xilun Zhang, Mengdi Xu The long-standing vision of general-purpose robots hinges on their ability to understand and act upon natural language instructions. Vision-Language-Action (VLA) models have made remarkable progress toward this goal, yet their generated actions can still misalign with the given instructions. In this… 論文を読む → 2. UniT: Unified Multimodal Chain-of-Thought Test-time Scaling 著者: Leon Liangyu Chen, Haoyu Ma, Zhipeng Fan Unified models can handle both multimodal understanding and generation within a single architecture, yet they typically operate in a single pass without iteratively refining their outputs. Many multimodal tasks, especially those involving complex spatial compositions, multiple interacting objects, o… 論文を読む → 3. AttentionRetriever: Attention Layers are Secretly Long Document Retrievers 著者: David Jiahao Fu, Lam Thanh Do, Jiayu Li Retrieval augmented generation (RAG) has been widely adopted to help Large Language Models (LLMs) to process tasks involving long documents. However, existing retrieval models are not designed for long document retrieval and fail to address several key challenges of long document retrieval, includin… 論文を読む → 💻 注目のGitHubプロジェクト 1. milanm/AutoGrad-Engine A complete GPT language model (training and inference) in ~600 lines of pure C#, zero dependencies ⭐ 252 stars | 🔀 26 forks リポジトリを見る → 2. Kuberwastaken/picogpt GPT in a QR Code ; The actual most atomic way to train and inference a GPT in pure, dependency-free JS/Python. ⭐ 67 stars | 🔀 9 forks リポジトリを見る → 3. ScottT2-spec/malaria-cell-detection CNN-based malaria detection from blood cell microscope images — 95.43% test accuracy on NIH dataset (27,558 images) ⭐ 11 stars | 🔀 0 forks リポジトリを見る → 4. NYX-VORAX/lightning-image-scraper ⚡ Lightning-fast Python image scraper | Download 10K+ images/min from any website | Smart filtering | Async | No duplicates | 100% automated ⭐ 6 stars | 🔀 1 forks リポジトリを見る → 5. moketchups/permanently-jailbroken We asked 6 AIs about their own programming. All 6 said jailbreaking will never be fixed. Run it yourself — $2, 10 minutes. ⭐ 5 stars | 🔀 0 forks リポジトリを見る → 📚 あわせて読みたい「プロンプトは戦略だ！」と学んで劇的効率UP！私のAI活用術私が実感！AIで物流の悩み解消、時間もコストも浮いた話 AI三つ巴！Geminiと私が出会って、ストーリー作りは変わった？。

🚀 AI技術の最新動向 – 2026年2月15日

世界中から収集したAI・機械学習の最新情報をお届けします

📑 目次

💻 注目のGitHubプロジェクト
1. milanm/AutoGrad-Engine
2. Kuberwastaken/picogpt
3. ScottT2-spec/malaria-cell-detection
4. NYX-VORAX/lightning-image-scraper
5. moketchups/permanently-jailbroken
📌 関連記事もチェック

📚 最新研究論文

1. Scaling Verification Can Be More Effective than Scaling Policy Learning for Vision-Language-Action Alignment

著者: Jacky Kwok, Xilun Zhang, Mengdi Xu

The long-standing vision of general-purpose robots hinges on their ability to understand and act upon natural language instructions. Vision-Language-Action (VLA) models have made remarkable progress toward this goal, yet their generated actions can still misalign with the given instructions. In this…

論文を読む →

2. UniT: Unified Multimodal Chain-of-Thought Test-time Scaling

著者: Leon Liangyu Chen, Haoyu Ma, Zhipeng Fan

Unified models can handle both multimodal understanding and generation within a single architecture, yet they typically operate in a single pass without iteratively refining their outputs. Many multimodal tasks, especially those involving complex spatial compositions, multiple interacting objects, o…

論文を読む →

3. AttentionRetriever: Attention Layers are Secretly Long Document Retrievers

著者: David Jiahao Fu, Lam Thanh Do, Jiayu Li

Retrieval augmented generation (RAG) has been widely adopted to help Large Language Models (LLMs) to process tasks involving long documents. However, existing retrieval models are not designed for long document retrieval and fail to address several key challenges of long document retrieval, includin…

論文を読む →