‼️🚨 UPDATE: The TanStack npm attack is now a full campaign.
'Mini' Shai-Hulud has hit:
- OpenSearch
- Mistral AI
- Guardrails AI
-UiPath
- Squawk packages across npm and PyPI
The malware specifically targets AI developer tooling. It hooks into Claude Code (.claude/settings.json) and VS Code (.vscode/tasks.json) to re-execute on every tool event, long after the infected package is gone. npm uninstall does not fix this.
We’ve identified a security incident that involved unauthorized access to certain internal Vercel systems, impacting a limited subset of customers. Please see our security bulletin:
https://t.co/0S939n3qHC
We’ve identified a security incident that involved unauthorized access to certain internal Vercel systems, impacting a limited subset of customers. Please see our security bulletin:
https://t.co/0S939n3qHC
EverMind has introduced Memory Sparse Attention (MSA), a sparse attention mechanism that integrates long-term memory directly into transformers. It supports contexts up to 100M tokens on standard hardware (2× A800 GPUs) with under 9% degradation on QA benchmarks from 16K to 100M tokens.
Key innovations include end-to-end differentiable compression/retrieval, document-wise RoPE, and memory interleaving for improved multi-hop reasoning. A 4B model outperforms larger RAG baselines in long-context QA.
This represents meaningful progress toward native long-context capabilities in LLMs, reducing reliance on external retrieval systems. Paper available now; code and model weights planned for open-source release.
Thoughts on how MSA compares to other long-context approaches like Ring Attention or Infini-Transformer?
#AILongContext #TransformerArchitectures #MachineLearning
How have I gone this long without knowing this website existed???
It literally displays all sorts of statistics, keeping track of our MPs and how their constituencies are doing...
This is amazing.
Releasing a new "Agentic Reviewer" for research papers. I started coding this as a weekend project, and @jyx_su made it much better.
I was inspired by a student who had a paper rejected 6 times over 3 years. Their feedback loop -- waiting ~6 months for feedback each time -- was painfully slow. We wanted to see if an agentic workflow can help researchers iterate faster.
When we trained the system on ICLR 2025 reviews and measured Spearman correlation (higher is better) on the test set:
- Correlation between two human reviewers: 0.41
- Correlation between AI and a human reviewer: 0.42
This suggests agentic reviewing is approaching human-level performance.
The agent grounds its feedback by searching arXiv, so it works best in fields like AI where research is freely published there. It’s an experimental tool, but I hope it helps you with your research.
Check it out here: https://t.co/n7ctnDilJJ
@ChrisMurphyCT You're being played by people who want regulatory capture.
They are scaring everyone with dubious studies so that open source models are regulated out of existence.
#hiring#malaysia#ai#ml#datascience
Ant International is needing some talented AI & Data Scientist, if this resonates and want to work in the world largest fintech, start applying here: https://t.co/6U8G7VcC4X
@Ant_Intl