GDP
@bookwormengr
Followers
10K
Following
57K
Media
1K
Statuses
48K
Artificial Intelligence @amazon, @awscloud Reinforcement Learning, OSS AI, General Purpose Agents, Autonomy All views personal!
Singapore, San Fransisco
Joined July 2010
@dhume Writing this as an Indian who works on AI in leadership role for one the largest companies in the world (though strictly my personal opinion, but based on verifiable data). You heard it first here: —————————- First some more shocks: You heard DeepSeek. Wait till you hear about
222
2K
6K
Interesting change of strategy by GLM4.7 compared to Kimi K2 Thinking, DeepSeek V3.2 and MiniMax M2.1 Interleaved thinking between tool calls: All these models supported interleaved thinking for tool calls, but they clear thinking from the previous turns as can be seen in the
GLM-4.7 further refines Interleaved Thinking and introduces Preserved Thinking and Turn-level Thinking. By enabling thought between actions and maintaining consistency across turns, it makes complex tasks more stable and controllable. https://t.co/6ZLi6Y9i7k
2
1
50
The coming years will split companies into two groups: those that become AI-native and those left catching up. The innovators who harness AI now will define the winners of tomorrow. Follow me for strategic insights on leading through the AI revolution.
16
21
229
2026 - Era of On Policy Distillation @rasbt
The LLM training eras: 202x Pre-training (foundation) 2022 RLHF + PPO 2023 LoRA SFT 2024 Mid-Training 2025 RLVR + GRPO
0
0
5
Just wow - "wild efficiency gains" with Thinking Machine's approach 🚀🚀🚀 "Post-train with MOPD: We adopted On-Policy-Distillation from Thinking Machine to merge multiple RL models, and the efficiency gains were wild. We matched the teacher model's performance using less than
MiMo-V2-Flash is live. It’s just step 2 on our AGI roadmap, but I wanted to dump some notes on the engineering choices that actually moved the needle. Architecture: We settled on a Hybrid SWA. It’s simple, elegant, and in our internal benchmarks, it outperformed other Linear
1
0
1
This is going to be multi-billion revenue toy. Movies could become a side income for Disney, if this happens.
Every single kid in the world will want these robots in future christmases Disney is solidifying themselves as leaders in the field of robotics. This is just the beginning.
0
0
2
Step out in high style at your next formal event and turn some heads with Gentleman's Guru. Shop the latest styles in men's formal wear and accessories for the modern gentleman. Get FREE shipping and EASY returns Get assistance from our Experts
0
24
153
Eliezer Yudkowsky, time to move to India. What will save us from skynet AGI is a power outage! Incompetent power grid people, will put AGI in its places. That means India is very safe from Skynet AGI.
This is insane, Waymo vehicles all over SF are bricked due to a power outage.
0
0
0
China is making its own America + Singapore.
This is largely being ignored but it's easily one of the biggest China news of the year. What China is doing with Hainan - a huge island (50 times the size of Singapore!) - is pretty extraordinary: they're basically making it into a completely different jurisdiction from the
1
0
5
Congrats Pieter Abbeel @pabbeel - Amazon's new AGI Head. TPOT needs no introduction to him. He is a pioneer in Robotics and DeepRL. Has 231K citations. He is Prof UC Berkeley; has been an Amazon Distinguished Scientist / VP / Scholar since 2024 focussed on Advancing AI and
29
90
2K
Wow, this is beautiful 💕
Today, the NeurIPS Foundation is proud to announce a $500,000 donation to OpenReview supporting the infrastructure that makes modern ML research possible. OpenReview has been our trusted partner for years, enabling rigorous peer review at the scale and pace our field demands.
0
0
3
Interesting example, @jsensarma . Converted it to a podcast with Notebook LM. In awe of the tool https://t.co/HQn2aI6V8E
@johnrushx - fyi
notebooklm.google.com
Use the power of AI for quick summarization and note taking, NotebookLM is your powerful virtual research assistant rooted in information you can trust.
Very interesting biological analog here. When chlorophyll first evolved, it led to a collapse of the existing biological ecosystem, before recovering slowly. Even though it was far superior way to harness energy. https://t.co/cExzbCnYNP
0
0
4
Can confirm what @tszzl says. For all its faults, USA is the most accepting society. There isn't another parallel. Also, the parents of the little girl need to prepare the girl for the world. You can not change everything about yourself - that lesson has to be taught with
@HinduHate @deedydas @ashtilawat america is the least racist country in the history of the world. you don’t prove anything with random emotional blackmail videos. I grew up here and people bend over backwards to accommodate different cultures and races
0
0
8
India too has mid-day meals scheme, but the food is low on protein. It is as if Indian policy makers hate protein.
Xi Jinping did one thing right: he guaranteed every schoolchild can get hot meals in schools since he came to power in 2013. Undernourishment, especially in early childhood, is strongly linked to lower IQ scores and impaired cognitive development.
5
1
26
New Turing test dropped. Why Boris won't use OpenAI Operator and integration with so many apps integrated? Not possible today, but some day.
Hiring (SF, hybrid): Part-time Personal Assistant. Looking for a high-judgment operator to run wide-ranging personal + household logistics end-to-end: vendors/services, errands, scheduling, research, bookings, travel/weekend plans, gifting, and “make life run smoothly”
1
0
8
USA has massive subsidy programs across industries, so does EU. Chinas is blamed uniquely though.
0
1
3
Beautiful post with one major issue - @levelsio Chinese government has given 230 B USD subsidy to entire EV industry over last 10 year. Peter points it as given them unfair advantage (though he doesn't complain about it. He is very objective.). However, that is 23B USD a year,
🇨🇳 Chinese EVs have been taking the world by storm, whether you like it or not When I look around here on the street in China I try recognize the car brands and I see so many brands I don't know One brand I see a lot is LI, but also BYD, Zeekr and XPeng Apparently there's now
1
2
2
List of China's underwhelming AI innovations is long 🤣 - MLA - DSA - GRPO - DAPO - BAGEL - Many SDE and ODE solver that power image/video generation They are so underwhelming, every major EU and US labs have adopted them. Losers.
🇨🇳 Compared to China "Algorithmically, I think the West still has an edge. I think none of the Chinese models or companies have shown that they can innovate on algorithmically something completely new thats beyond the state of the art. They have been very good at following the
0
0
7
Wow, DeepSeek 3.2 Speciale #2 for advanced physics reasoning. Didn't expect it to be that high.
GPT-5.2 xhigh by @OpenAI scores 0% (zero) on CritPt. CritPt is a research-level physics reasoning benchmark designed to test whether an AI can do genuine, expert-grade theoretical physics reasoning. The highest score is 9.1% by Gemini 3 Pro.
0
0
3
Homeschooling Resources, Workshops & more! Attendees from 38 states!
0
46
222