📝 この記事のポイント
- 🚀 AI技術の最新動向 - 2026年3月7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hanxiao/mlx-visSamYbarra/Polymarket-AI-Trading-Botmechramc/Oriongeorgeguimaraes/hallmarkErdemYavuz55/drug-review-sentiment-analysis📌 関連記事もチェック📚 最新研究論文1. RoboPocket: Improve Robot Policies Instantly with Your Phone著者: Junjie Fang, Wendi Chen, Han XueScaling imitation learning is fundamentally constrained by the efficiency of data collection. While handheld interfaces have emerged as a scalable solution for in-the-wild data acquisition, they predominantly operate in an open-loop manner: operators blindly collect demonstrations without knowing th...論文を読む →2. POET-X: Memory-efficient LLM Training by Scaling Orthogonal Transformation著者: Zeju Qiu, Lixin Liu, Adrian WellerEfficient and stable training of large language models (LLMs) remains a core challenge in modern machine learning systems. To address this challenge, Reparameterized Orthogonal Equivalence Training (POET), a spectrum-preserving framework that optimizes each weight matrix through orthogonal equivalen...論文を読む →3. The Spike, the Sparse and the Sink: Anatomy of Massive Activations and Attention Sinks著者: Shangwen Sun, Alfredo Canziani, Yann LeCunWe study two recurring phenomena in Transformer language models: massive activations, in which a small number of tokens exhibit extreme outliers in a few channels, and attention sinks, in which certain tokens attract disproportionate attention mass regardless of semantic relevance. Prior work observ...論文を読む →💻 注目のGitHubプロジェクト1. hanxiao/mlx-visPure MLX implementations of UMAP, t-SNE, PaCMAP, TriMap, DREAMS, CNE, and NNDescent for Apple Silicon. Metal GPU for computation and video rendering.⭐ 53 stars | 🔀 0 forksリポジトリを見る →2. SamYbarra/Polymarket-AI-Trading-BotPolymarket trading bot based on AI | ML⭐ 21 stars | 🔀 21 forksリポジトリを見る →3. mechramc/OrionLocal AI runtime for training & running small LLMs directly on Apple Neural Engine (ANE). No CoreML. No Metal. Offline, on-device fine-tuning & inference on M-series silicon.⭐ 14 stars | 🔀 1 forksリポジトリを見る →4. georgeguimaraes/hallmarkHallucination detection for Elixir, powered by Vectara's HHEM model⭐ 10 stars | 🔀 0 forksリポジトリを見る →5. ErdemYavuz55/drug-review-sentiment-analysisEnd-to-end sentiment analysis pipeline on 160K+ drug reviews using TF-IDF, Word2Vec, and fine-tuned BERT for binary and multi-class classification.⭐ 9 stars | 🔀 0 forksリポジトリを見る →。
🚀 AI技術の最新動向 – 2026年3月7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
- hanxiao/mlx-vis
- SamYbarra/Polymarket-AI-Trading-Bot
- mechramc/Orion
- georgeguimaraes/hallmark
- ErdemYavuz55/drug-review-sentiment-analysis
- 📌 関連記事もチェック
📚 最新研究論文
1. RoboPocket: Improve Robot Policies Instantly with Your Phone
著者: Junjie Fang, Wendi Chen, Han Xue
Scaling imitation learning is fundamentally constrained by the efficiency of data collection. While handheld interfaces have emerged as a scalable solution for in-the-wild data acquisition, they predominantly operate in an open-loop manner: operators blindly collect demonstrations without knowing th…
2. POET-X: Memory-efficient LLM Training by Scaling Orthogonal Transformation
著者: Zeju Qiu, Lixin Liu, Adrian Weller
Efficient and stable training of large language models (LLMs) remains a core challenge in modern machine learning systems. To address this challenge, Reparameterized Orthogonal Equivalence Training (POET), a spectrum-preserving framework that optimizes each weight matrix through orthogonal equivalen…
3. The Spike, the Sparse and the Sink: Anatomy of Massive Activations and Attention Sinks
著者: Shangwen Sun, Alfredo Canziani, Yann LeCun
We study two recurring phenomena in Transformer language models: massive activations, in which a small number of tokens exhibit extreme outliers in a few channels, and attention sinks, in which certain tokens attract disproportionate attention mass regardless of semantic relevance. Prior work observ…
💻 注目のGitHubプロジェクト
1. hanxiao/mlx-vis
Pure MLX implementations of UMAP, t-SNE, PaCMAP, TriMap, DREAMS, CNE, and NNDescent for Apple Silicon. Metal GPU for computation and video rendering.
⭐ 53 stars | 🔀 0 forks
2. SamYbarra/Polymarket-AI-Trading-Bot
Polymarket trading bot based on AI | ML
⭐ 21 stars | 🔀 21 forks
3. mechramc/Orion
Local AI runtime for training & running small LLMs directly on Apple Neural Engine (ANE). No CoreML. No Metal. Offline, on-device fine-tuning & inference on M-series silicon.
⭐ 14 stars | 🔀 1 forks
4. georgeguimaraes/hallmark
Hallucination detection for Elixir, powered by Vectara's HHEM model
⭐ 10 stars | 🔀 0 forks
5. ErdemYavuz55/drug-review-sentiment-analysis
End-to-end sentiment analysis pipeline on 160K+ drug reviews using TF-IDF, Word2Vec, and fine-tuned BERT for binary and multi-class classification.
⭐ 9 stars | 🔀 0 forks
