X-Tokenizer has to be one of the hardest papers ive studied so far, it has so many advancements than regular FAST or VQ-VAE methods, will share my summary and learnings soon.
India’s Longest 24 Hour Physical AI Hackathon, co-hosted by https://t.co/ECiAlGZu4M and @xprobotics_ai .
Hello All!
This October, https://t.co/ECiAlGZu4M is bringing builders, roboticists, AI engineers, researchers, students and deeptech founders together for a series of intense 24-hour AI Hackathons in Bengaluru.
B1’s Hacktober is built around one simple idea
The next generation of AI will not remain confined to screens. It will take up different modalities, perceive, reason, manipulate, move and operate in the physical world.
On 17 October 2026, teams of up to four will come together at https://t.co/ECiAlGZu4M to build hands-on Physical AI systems using robotic hardware, embodied intelligence workflows and modern AI tooling.
Every Hackathon will begin with a dedicated Workshop, this one on Physical AI, delivered by former physical-AI and robotics researchers @nvidia , @CarnegieMellon and @iiscbangalore
The workshop will help participants orient themselves on how modern robotic intelligence is being built across perception, manipulation, control, simulation and learning-based systems.
Participants will get access to LeRobot SO-101 robotic arms, NVIDIA Brev Credits, technical mentorship and an environment designed specifically for experimentation. We want teams to move quickly from an idea to something that can actually make a dent in the World!
The Top 3 teams will win a pair of LeRobot Arms, enabling them to continue building beyond Hacktober. Selected winning teams will also get the opportunity to pitch and demonstrate their work live at the AI & Robotics Pavilion at Bengaluru Tech Summit 2026, putting their work in front of a much larger technology, industry and innovation ecosystem.
You do not need to arrive with a finished product. You need curiosity, engineering instinct and the willingness to build.
B1’s Hacktober — Physical AI 2026
📅 17 October 2026
⏱️ 24 Hours
📍 https://t.co/ECiAlGZu4M, Bengaluru
👥 Teams of up to 4
💰 ₹5,000 per team
🏆 Top 3 teams win a LeRobot Arm
🚀 Winning teams pitch live at the AI & Robotics Pavilion, @blrtechsummit 2026
We are grateful to the generous support extended from Get Set Robotics, Tanna TechBiz LLP , Solo Tech and RIC (Robotics India Community).
If you have been waiting for an excuse to stop watching Physical AI from the sidelines and start building it, this is it.
Builders, researchers, roboticists, founders and students, we’re ready for you.
Register here:
https://t.co/GGxRwNhde5
More Details and Registration Details
https://t.co/3Wm7fYvsvz?
Unpopular opinion: You don’t need a new productivity app.
You just need F11.
Using Windows in full screen mode eliminates 70% of visual distractions instantly. It’s the most underrated focus hack in existence.
Argus: an open-source robotics data annotation and quality pipeline.
With new frontier VLMs like GPT-6 Astra, we can generate rich, high-quality annotations for robotics data for both training and dataset analysis. Argus delivers detailed, timestamp-level annotations while catching issues like mislabeled instructions, sped-up recordings, swapped camera streams, and unflagged operator mistakes.
To support better data quality for the robotics community, we’re open-sourcing Argus for anyone to use 🧵
created a 'trackable html file' for my October's study plan, been loving this way of tracking my goals lately.
ive split my robot learning study into theory as well as code implementation chunks so that i dont just read read read but also implement it in coding as much as i can!
JUST IN: Google reveals Gemini 4 Argon has a 1 million-token output window, nearly 700% larger than leading frontier models from OpenAI & Anthropic.