# May 18, 2025

Daily

Weekly

- **Spaced Repetition Systems (SRS) have evolved significantly** with the introduction of the **FSRS algorithm**, which utilizes machine learning to optimize review intervals based on individual recall probabilities, enhancing learning efficiency.

- **AniSora** is an **open-source animated video generation model** by Bilibili, enabling one-click creation of high-quality anime videos from images or prompts, tailored for diverse styles like VTuber content and manga adaptations.

- **LLM persuaders** (Claude Sonnet 3.5) outperformed incentivized human persuaders in a large-scale experiment, achieving higher compliance in both **truthful** and **deceptive** contexts, indicating a significant leap in AI's persuasive capabilities. [Link to article](https://arxiv.org/abs/2505.09662)

- The **Voynich Manuscript** analysis reveals that it exhibits **structured language-like behavior**, suggesting it may encode a constructed or mnemonic language, despite remaining undeciphered.

- This study reveals that **N-gram statistics** can effectively approximate transformer predictions, leading to insights such as a method for detecting **overfitting** during training without a holdout set.

- The **Armv9 architecture** has significantly boosted Arm Holdings' financial performance, achieving over **$1 billion** in quarterly revenue and **$4 billion** in annual sales for the first time, driven by higher royalty fees and increased adoption by hyperscalers.

- The **SparseDepthTransformer** innovatively routes tokens through a variable number of layers based on their **semantic importance**, enhancing efficiency in transformer models.

- **AlphaEvolve** is a coding agent that utilizes **evolutionary algorithms** to autonomously develop programs, leveraging the vast computing power of consumer-grade GPUs, which can reach nearly **50 TeraFLOPS** in performance.

- **Pivotal Token Search (PTS)** optimizes language model training by identifying **critical decision points** where specific tokens significantly influence the outcome, enhancing efficiency and effectiveness in model performance.

- **cachelm – Semantic Caching for LLMs (Cut Costs, Boost Speed)**

- Inspired by Anthropic’s Biology of an LLM: Exploring Prompt Cues in Two LLMs