Links for 2024-06-08
AI:
Eric Schmidt poached talent from Apple, SpaceX, and Google to create AI military drones for Ukraine. https://www.forbes.com/sites/sarahemerson/2024/06/06/eric-schmidt-is-secretly-testing-ai-military-drones-in-a-wealthy-silicon-valley-suburb/ [No paywall: https://archive.is/lkr04]
Scott Aaronson recommends Leopold Aschenbrenner's essay: "With unusual clarity, concreteness, and seriousness...Leopold sets out his vision of how AI is going to transform civilization over the next 5-10 years." https://scottaaronson.blog/?p=8047
Will jailbreaking soon be a solved issue? “We introduce Short Circuiting: the first alignment technique that is adversarially robust. Unlike adversarial training which takes days, short circuits can be inserted in under 20 minutes on a GPU. Unlike input/output filters, short circuited models are deployed as normal models with no additional inference cost.” https://arxiv.org/abs/2406.04313
Will We Run Out of Data? Limits of LLM Scaling Based on Human-Generated Data https://epochai.org/blog/will-we-run-out-of-data-limits-of-llm-scaling-based-on-human-generated-data
Self-Improving Robust Preference Optimization — “…we derive a practical, but mathematically principled offline algorithm to explicitly teach a model to self-improve and be robust to the choice of the eval task at the same-time!” https://arxiv.org/abs/2406.01660
MatMul-free LLMs: Proposes an implementation that eliminates matrix multiplication operations from LLMs while maintaining performance at billion-parameter scales. https://arxiv.org/abs/2406.02528
Grokfast: significantly reduces training iterations, accelerating the grokking process by 50 times in machine learning models. https://arxiv.org/abs/2405.20233
Buffer of Thoughts: Significant performance improvements over previous SOTA methods: 11% on Game of 24, 20% on Geometric Shapes and 51% on Checkmate-in-One. https://github.com/YangLing0818/buffer-of-thought-llm
BitsFusion: Compresses the UNet of Stable Diffusion v1.5 (1.72 GB, FP16) into 1.99 bits (219 MB), achieving a 7.9X compression ratio and even better performance. https://snap-research.github.io/BitsFusion/
YOLOv10, a powerful real-time object detection model, reduces latency by 46% and parameter count by 25% compared to its predecessor. https://github.com/THU-MIG/yolov10
σ-GPTs: A New Approach to Autoregressive Models https://arxiv.org/abs/2404.09562
Google releases new tool to automate Python code optimization. https://labs.google.com/code/transformer
Miscellaneous:
“Using the strategy game Civilization, this proof-of-concept study explores if strategy video games are indicative of managerial skills and, if so, of what managerial skills…We find that students who had high scores in the game had better skills related to problem-solving and organizing and planning than the students who had low scores.” https://link.springer.com/article/10.1007/s11846-020-00378-0
Ukraine:
🤯🔥 🇺🇸US-supplied M2A2 Bradley of the Ukrainian 47th Mechanised Brigade destroys 🇷🇺Russian BTR-82A in super close range with its 25mm Bushmaster cannon https://x.com/GloOouD/status/1799113513270653037
92nd brigade at work, pounding russian positions in Hlyboke. the spot where it drives is very interesting, the devastating effects of it's 30mm cannon. https://x.com/Teoyaomiquu/status/1799173430371119588
Dangerously close DPICM. Wounded Russians, FPVs finishing them off. https://x.com/DefMon3/status/1799174117490339985
The moment of the strike of one of the drones on Russian airbase in Mozdok https://x.com/bayraktar_1love/status/1799407565430006119
Yak-52, which is used to shoot down Russian drones in the lens of the Russian Zala drone over the Mykolaiv region. https://x.com/GloOouD/status/1799390808254140692
Russians on motorbikes tried to storm positions of the 425th separate assault battalion ‘Skala’ but something went wrong. https://x.com/NOELreports/status/1799408470359429375
Russian state TV shows the destroyed Ukrainian town of Vovchansk and threatens that German cities will look the same https://x.com/den_kazansky/status/1799046473528819848