OMG!
neuralmagic | 11/22/23 Key Takeaways We expanded our Sparse Fine-Tuning research results to include Llama 2. The results include 60% sparsity with INT8 quantization and no drop in accuracy. DeepSparse now supports accelerated inference of sparse-quantized Llama 2 models, with inference speeds 6-8x faster over the baseline at 60-80% sparsity. We used some interesting algorithmic …
Elon Musk needed fewer than 100 characters to add new chaos to the ongoing crisis swirling around OpenAI after the shock firing of CEO Sam Altman last week. In a post on X on Tuesday, Musk drew attention to an anonymous letter accusing Altman of various examples of underhanded behavior as CEO of OpenAI. The …
Wang Huning, fourth-ranking member of the Politburo Standing Committee, has served as China’s ideologue-in-chief for the past two decades. He has been called the “world’s most dangerous thinker” and the “éminence grise” of Xi Jinping. Yet long before Wang was elevated to high office, he was a young political scientist eager to learn everything he …
It started about a week ago. I use Firefox, locked down a bit, with UBlock Origin. Anyone else getting issues? I recently switched ISP and it happened after that but it also happens with NordVPN running. Nothing else changed with my setup. This is getting ridiculous. I’ve read on here many many times about Cloudflare …
The 30-hour workweek: Often celebrated by employees, feared by business representatives. Discussions about a shorter workweek have been ongoing for a long time and don’t seem to be ending. What future scenarios can arise from changes in working hours, and is it realistic that we will all only work three hours a day in the …
Credit: internet At Alan, we’ve always been committed to providing the best user experience possible, and this extends to our own engineers, through what we call the developer experience. We believe that a great developer experience is essential for building high-quality software and that it’s a key factor in attracting and retaining top talent. We’ve …
The Free Software Foundation (FSF) is the steward of the family of GNU General Public Licenses (GPL), which we designed with care to ensure that copyright holders have a tool to release programs as free software in a way that perpetuates the freedom of computer users. The licenses include explicit grant of rights necessary to …
Low-rank adaptation (LoRA) is among the most widely used and effective techniques for efficiently training custom LLMs. For those interested in open-source LLMs, it’s an essential technique worth familiarizing oneself with. Last month, I shared an article with several LoRA experiments, based on the open-source Lit-GPT repository that I co-maintain with my colleagues at Lightning …