Mirza Mushtaq Baig

Espresso aficionado ☕| Founder: KineticAI Labs | Robotics | Physical AI | Quantum Computing | OpenAI Forum member | Comic book geek

[Mirza Mushtaq Baig @MirzaMB]

To the unsung heroes of the @USDS @NIST & @NSF who have been terminated or laid off - if you want to work on the challenges facing the field of AI Robotics and have experience in #AI / #ML / #Quantum or #Robotics, do connect with me: mirza(at)https://t.co/2aR4oH91pe

[Mirza Mushtaq Baig @MirzaMB]

2) Humanoid robots consume significant power for actuators (think the muscles of a robot), what's needed then? High-torque electric actuators. Tactile sensors with sub-millimeter precision. Modular joints for dynamic balance.

Sound challenging? Sure is.

[Mirza Mushtaq Baig @MirzaMB]

1) lack the precision needed for robotic tasks like real-time environmental perception and object manipulation. Domain-specific AI models are required.

Data & Controls Engineers must together build tools to identify anomalies in sensor data, joint movements, task performance

[Mirza Mushtaq Baig @MirzaMB]

4) A strong America requires (to borrow from Dan Goldin, NASA's 9th Chief), 'courageous collaboration, and a widespread commitment to [solving] hard problems' not a wholesale 'deleting' of the good men and women who have served this nation's interests and principles.

[Mirza Mushtaq Baig @MirzaMB]

3) ... that if you have an open role right now, make sure to consider these unsung heroes: now is the time we pay our dues and uplift those who have been let down.

Remember: service to others is the rent you pay for your room here on Earth.

[Mirza Mushtaq Baig @MirzaMB]

2) This is a reminder to those federal workers of the Latin saying 'ad astra per aspera,' to the stars through hardships.

This is also an open call, to my peers, colleagues, and cohorts in the enterprise and startup space across the United States

[Mirza Mushtaq Baig @MirzaMB]

1) My heart goes out to all the federal workers right now - you inherently understand the principle of service before self. A lifetime ago, I worked with several U.S. agencies and know firsthand, the good, hardworking, smart, and dedicated men and women who serve the government

[Mirza Mushtaq Baig @MirzaMB]

@DarrelFrater let's move the convo to LinkedIn - just sent you a connection request

[Mirza Mushtaq Baig @MirzaMB]

Happy to share that I'll be speaking at Use-Case: The Impact of Quantum Computing in AI! Make sure to attend it on February 10. https://t.co/hVmqrVKGEh

#QuantumAI #QuantumComputing #AI

[Sophia Yang @sophiamyang]

🚀 Announcing the all new @MistralAI Le Chat – your ultimate AI sidekick for life & work, now live on mobile!

So many exciting features 🧵:

[Lior Alexander @LiorOnAI]

Wow, someone just released a notebook to train a reasoning LLM with the new RL algorithm from DeepSeek, GRPO.

In <2 hours, you can transform a very small model, Qwen 0.5 (500 million parameters) into a tiny math reasoning machine.

[AK @_akhaliq]

Google just dropped Gemini 2.0 for everyone

lets test it out with coder mode

prompt: write a script for 100 bouncing bright yellow balls within a sphere, make sure to handle collision detection properly. make the sphere slowly rotate. make sure balls stays within the sphere. implement it in p5.js

[Mirza Mushtaq Baig @MirzaMB]

Thanks @edgemiddleeast Edge ME

https://t.co/6LRmLUtCRD

[Mirza Mushtaq Baig @MirzaMB]

Thanks to @stockhouse team

https://t.co/CNHPJrG9Wd

[Jim Fan @DrJimFan]

We trained a robot dog to balance and walk on top of a yoga ball purely in simulation, and then transfer zero-shot to the real world. No fine-tuning. Just works.

I’m excited to announce DrEureka, an LLM agent that writes code to train robot skills in simulation, and writes more code to bridge the difficult simulation-reality gap. It fully automates the pipeline from new skill learning to real-world deployment.

The Yoga ball task is particularly hard because it is not possible to accurately simulate the bouncy ball surface. Yet DrEureka has no trouble searching over a vast space of sim-to-real configurations, and enables the dog to steer the ball on various terrains, even walking sideways!

Traditionally, the sim-to-real transfer is achieved by domain randomization, a tedious process that requires expert human roboticists to stare at every parameter and adjust by hand. Frontier LLMs like GPT-4 have tons of built-in physical intuition for friction, damping, stiffness, gravity, etc. We are (mildly) surprised to find that DrEureka can tune these parameters competently and explain its reasoning well.

DrEureka builds on our prior work Eureka, the algorithm that teaches a 5-finger robot hand to do pen spinning. It takes one step further on our quest to automate the entire robot learning pipeline by an AI agent system. One model that outputs strings will supervise another model that outputs torque control.

We open-source everything! Welcome you all to check out the paper, more videos, and try the codebase today: https://t.co/RwiBT3z78H
Code: https://t.co/ERp4Gl0N36

[Vaskange @Vaskange]

The original video of my artwork here.
Stay tuned, to discover more infinite stories!

[DIA Oracles @DIAdata_org]

🎙 Panel on NFTs, DeFi and Oracles

Join us on a round table discussion about how #oracles are powering #NFTfi use cases.

👥 with:
@pclaudius (DIA)
@jamie247(@OVioHQ)
@moonsman_sy (@LabsSumeria)

🗓️ Tue, Jul 26 @ 17:00 BST
📍https://t.co/vV84VqJZqS

[Mirza Mushtaq Baig @MirzaMB]

@NFL Expands Ticket Stub #NFTs to Over 100 Games for 2022 Season

https://t.co/gc3iU5Yk2C

[Mirza Mushtaq Baig @MirzaMB]

🤯 Croatian #Web3 startup offers lifetime @Netflix and @Spotify membership via NFTs

https://t.co/FzZUaK5eF1