The RLHF method behind the best open models! Both @deepseek_ai and @Alibaba_Qwen use GRPO in post-training! Group Relative Policy Optimization. GRPO was introduced in the DeepSeekMath Paper last year to improve mathematical reasoning capabilities with less memory consumption,…
✨🎨 Edit Pro AI: Unleash limitless creativity! ✨🎨
🔮 Transform photos & videos instantly
🚀 Harness boundless AI power
🖼️ From basic edits to mind-blowing effects
Experience the magic - Try it now! 🌟
#EditProAI#AImagic#CreateWithoutLimits
KANs (NNs with learned functions on the edges) have a quite elegant representation using Tensor Diagrams.
This chart of MLP layers also shows some neat relationship between things like ReGLUs and MoEs.
MambaMixer
Efficient Selective State Space Models with Dual Token and Channel Selection
Recent advances in deep learning have mainly relied on Transformers due to their data dependency and ability to learn at scale. The attention module in these architectures, however,
How to teach a language model a new language without retraining?
The key: periodic forgetting.
A recent research paper demonstrated how forgetting can enhance the plasticity of a language model. 👇🏼
1/8
* Language is low bandwidth: less than 12 bytes/second. A person can read 270 words/minutes, or 4.5 words/second, which is 12 bytes/s (assuming 2 bytes per token and 0.75 words per token). A modern LLM is typically trained with 1x10^13 two-byte tokens, which is 2x10^13 bytes.…
* Language is low bandwidth: less than 12 bytes/second. A person can read 270 words/minutes, or 4.5 words/second, which is 12 bytes/s (assuming 2 bytes per token and 0.75 words per token). A modern LLM is typically trained with 1x10^13 two-byte tokens, which is 2x10^13 bytes.…
Beautiful work / attention to detail trying to get Gemma to finetune correctly. There are so many foot guns here to be super careful with. All of these issues don't throw any errors, they silently make your network worse.
A great example of what I wrote about in my "A Recipe for…
Beautiful work / attention to detail trying to get Gemma to finetune correctly. There are so many foot guns here to be super careful with. All of these issues don't throw any errors, they silently make your network worse.
A great example of what I wrote about in my "A Recipe for…
Microsoft launched the best course on Generative AI!
The free 18 lesson course is available on Github and will teach you everything you need to know to start building Generative AI applications.
293 Followers 1K FollowingAdroit Ignite HMI Software is optimised for Windows and built on the best technologies availablewhich makes it a more flexible, simpler, smarter and faster.
16 Followers 405 Following🤖 AI enthusiast | 📚 Bookworm | 🌍 Traveler with a penchant for tech | 🍕 Pizza lover | 🎧 Music explorer |
Balancing life between BYTES & Adventures !
5K Followers 4K FollowingWelcome to https://t.co/9uFCHRnq0d - A new DeFi tool that allows users to create and perform a Flash loan backed trade from an easy to use UI.
17K Followers 1K Following专注 - Context Engineering, AI (Coding) Agents, RAG etc.
分享 - AI papers, news, apps and OSS.
ex Microsoft MVP (2014-2022)
📢 公众号/小红书: AI 启蒙小伙伴
🔗 信息卡提示词 🔽
17K Followers 574 FollowingWe make AI models Dolphin and Samantha
BTC 3ENBV6zdwyqieAXzZP2i3EjeZtVwEmAuo4
https://t.co/3ri2GbWU13
https://t.co/zH0F3pSLuq @dphnAI
83K Followers 8K FollowingCompiling in real-time, the race towards AGI.
🗞️ Don't miss my daily top 1% AI analysis newsletter directly to your inbox 👉 https://t.co/6LBxO8215l
39K Followers 263 FollowingI eye AI | Making the most of what AI has to offer | Always looking for the next big thing | Follow to keep an 👁️ on the latest Tools, Tutorials & Prompts.
3.1M Followers 150 FollowingEngineer. Selecting and curating pictures and videos trying to awaken your sense of wonder. Science, tech, art, weather, space, the unusual around us.
962 Followers 356 FollowingPh.D. student at Nagoya University, working on motion planning and control in the fields of robotics and autonomous driving.
7K Followers 323 FollowingChampioning open-source projects and high-quality, informative content related to robotics. Subscribe: https://t.co/IX1YhgfOkE
6K Followers 1K FollowingThe industry leaders in solving the hardest robotics problems.
We provide advanced solutions for structured and unstructured environments on Earth and in space.
153K Followers 5K FollowingSubscribe to my DeFi blog to get ahead of the curve 👉 https://t.co/7O0WAdXUnT
Co-founder of @PinkBrains_io DeFi Creator Studio
11K Followers 4 FollowingIBC is a blockchain interoperability protocol used by 100+ chains. It enables secure, permissionless, feature-rich cross-chain interactions.
10K Followers 427 Followingbuilding the everything wallet
founder of @blormmy.
Use blormmy to do swaps on X, buy over +1 b items from amazon, and mint tweets as NFTs.
19K Followers 739 FollowingHome of the annual Open Hardware Summit hybrid remote & inperson summit celebrating open source creations l Edinburgh 2025⚡⚙️🔧 News at @oshwassociation
2K Followers 4 FollowingWe make software for robots.
Applications: Mobile robots and forklifts for factories, floor cleaning, autonomous boats, warehouse fleets, lawn mowers and more!
5K Followers 561 FollowingYour hub for #PX4, #MAVSDK, #MAVLINK, #QGC community news and updates. Tweets by community managers @Dronecode Foundation. #opensource #drones #robotics
117K Followers 377 FollowingNVIDIA Robotics inspires visionaries and developers to create the next generation of AI-driven robots and explore the world of physical AI.
32K Followers 3K FollowingExploring the business and applications of robotics. https://t.co/LkUs02AGiV has a worldwide database and global map of thousands of companies in robotics.