#cplusplus
FlashMLA is DeepSeek's optimized attention library that makes AI models run faster and use less memory. It works with advanced NVIDIA GPUs to speed up how language models process information, achieving up to 660 trillion floating-point operations per second. The library supports both dense and sparse attention modes, meaning it can focus on important tokens while skipping less relevant ones, reducing computational waste. For you, this means faster AI responses, lower costs for running large language models, and better performance on tasks like chatbots and code generation. The technology is open-source and integrates with popular AI frameworks like PyTorch and Hugging Face, making it accessible for developers building next-generation AI applications.
https://github.com/deepseek-ai/FlashMLA
LinkWarden
Self-hosted, open-source #bookmark + archive manager to collect, and save websites for offline use.
The objective is to have a self-hosted place to keep useful links in one place, and since useful links can go away (see the inevitability of Link Rot), LinkWarden also saves a copy of the link as screenshot and PDF.
https://github.com/Daniel31x13/link-warden
📖 Inside the First Art Gallery for Blind Artists and Audiences #bookmark#raindrop
https://www.thrillist.com/travel/nation/envision-arts-gallery-wichita-kansas