# Mar 24, 2026

## LiteLLM Python package compromised by supply-chain attack
- The **`litellm==1.82.8`** package on PyPI contains a **malicious `.pth` file** that executes a **credential-stealing script** upon Python interpreter startup, compromising sensitive data without requiring an import statement.

## Epoch confirms GPT5.4 Pro solved a frontier math open problem
- **A Ramsey-style problem on hypergraphs** seeks to construct large hypergraphs without a specific property, with a recent solution confirmed by AI models, notably GPT-5.4 Pro, which improved the understanding of lower bounds in hypergraph theory.

## LLM Neuroanatomy II: Modern LLM Hacking and Hints of a Universal Language?
- **RYS method** demonstrates that **relayering** in modern LLMs like Qwen3.5-27B enhances performance, confirming that this technique is not merely a quirk of earlier models but a robust feature of Transformer architecture.

## MSA: Memory Sparse Attention
- **Memory Sparse Attention (MSA)** offers a **scalable, end-to-end trainable framework** that efficiently handles **100M-token contexts**, achieving **<9% degradation** from **16K to 100M tokens** through innovative techniques like **sparse attention** and **document-wise RoPE**.

## Finding all regex matches has always been O(n²)
- **Regex engines** universally exhibit **O(n²)** complexity when finding all matches, despite claims of linear performance; this issue has persisted since the 1970s, affecting all languages and implementations.

## Show HN: Gemini can now natively embed video, so I built sub-second video search
- **SentrySearch enables semantic search over dashcam footage**, utilizing Google's Gemini Embedding model to convert video into a searchable vector space, allowing users to retrieve specific clips based on text queries.

## Arm AGI CPU
- **Arm AGI CPU** marks a pivotal advancement in AI infrastructure, delivering **breakthrough performance** and efficiency tailored for agentic AI workloads, enabling real-time decision-making across distributed systems.

## Run a 1T parameter model on a 32gb Mac by streaming tensors from NVMe
- **Hypura** is a **storage-tier-aware LLM inference scheduler** designed for Apple Silicon, enabling the execution of models larger than the system's memory by intelligently distributing tensors across GPU, RAM, and NVMe based on usage patterns and hardware capabilities.

## ARM AGI CPU: Specs and SKUs
- **ARM AGI CPU** is Arm's inaugural production silicon, engineered for **AI infrastructure** with up to **136 Neoverse V3 cores** and a **3nm process**, enhancing performance and density for modern data centers.

## Designing AI for Disruptive Science
- **Scaling AI does not guarantee paradigm shifts**; true innovation requires AI to generate new conceptual frameworks rather than merely enhancing existing predictive capabilities.

## Welcome to FastMCP
- **FastMCP is the premier framework for developing MCP applications**, enabling seamless integration of LLMs with tools and data while automating schema, validation, and documentation processes.

## [R] Causal self-attention as a probabilistic model over embeddings
- **Support tokens** emerge from a new interpretation of causal self-attention transformers, revealing a **stability-margin** analogous to support vector machines, which enhances the robustness of large language models (LLMs).

## [R] VLouvain: Louvain Community Detection Directly on Vectors, No Graph Construction
- **VLouvain** enables community detection directly on **embedding matrices**, eliminating the need for graph construction and reducing complexity from **O(n²)** to **O(n*d)**, thus maintaining performance even with large datasets.

## PyTorch 2.11 Release Blog
- **PyTorch 2.11 introduces significant enhancements** such as **Differentiable Collectives for Distributed Training** and **FlexAttention with FlashAttention-4**, which promise to improve performance and flexibility in deep learning workflows.

## [D] Modeling online discourse escalation as a state machine (dataset + labeling approach)
- The proposed framework models **online discourse escalation** as a **state machine**, identifying seven distinct states from **Neutral** to **Threats of violence**, with each comment labeled according to its local state while the thread evolves globally.