Something interesting.

The worminator project.

To reverse engineer an entire nervous system
https://arxiv.org/abs/2308.06578

Here we argue that the time is ripe for systems neuroscience to embark on a concerted effort to reverse engineer a smaller system and that Caenorhabditis elegans is the ideal candidate system as the established optophysiology techniques can capture and control each neuron’s activity and scale to hundreds of thousands of experiments. Data across populations and behaviors can be combined because across individuals the nervous system is largely conserved in form and function. Modern machine-learning based modeling should then enable a simulation of C. elegans’ impressive breadth of brain states and behaviors. The ability to reverse engineer an entire nervous system will benefit the design of artificial intelligence systems and all of systems neuroscience, enabling fundamental insights as well as new approaches for investigations of progressively larger nervous systems.

Лёгкое чтение на выходные

https://www.quantamagazine.org/the-useless-perspective-that-transformed-mathematics-20200609/

Давно мы Шмидхубера не постили!

https://www.youtube.com/watch?v=q27XMPm5wg8

Добрая вечная классика:

https://www.youtube.com/watch?v=C2vgICfQawE

И в дополнение для тех, кто соскучился, хорошая реализация с каталогом известных интересных конфигураций:

https://playgameoflife.com/

On LLM challenges

https://huyenchip.com/2023/08/16/llm-research-open-challenges.html

Hot news!
----------------------------

Today we’re announcing SeamlessM4T, the first all-in-one, multilingual multimodal AI translation model.

Details ➡️ https://bit.ly/45z0e6s
Demo ➡️ https://bit.ly/3YNwm3Z

This single model can perform tasks across speech-to-text, speech-to-speech, text-to-speech, text-to-text translation & speech recognition for up to 100 languages depending on the task. Compared to cascaded approaches, SeamlessM4T's single system reduces errors & delays, increasing translation efficiency and delivering state-of-the-art results.

As part of our open approach, we're publicly releasing this work under a CC BY-NC 4.0 license so that others can continue to build on this important field of study.

И ещё из хороших новостей: теперь можно файнтюнить gpt-3.5-turbo

https://openai.com/blog/gpt-3-5-turbo-fine-tuning-and-api-updates

Interesting numbers

More than 15 billion images created using text-to-image algorithms since last year. To put this in perspective, it took photographers 150 years, from the first photograph taken in 1826 until 1975, to reach the 15 billion mark.

https://journal.everypixel.com/ai-image-statistics