Tomas Ramanauskas

Machine Learning Enthusiast

[Tomas Ramanauskas @Tomas0413]

Many very interesting and clearly expressed thoughts!

[Richard Socher @RichardSocher]

We can safely advance science and society with AI or get distracted with scifi doom scenarios that require magical extrapolation. I really enjoyed this podcast on RSI for science.
https://t.co/B1gALzCzR7

[Lilian Weng @lilianweng]

new post on harness engineering for AI self-improvement: https://t.co/ZYvGfVs61k

It is hard to forecast how much the future of RSI will rely on harnesses. Likely harness engineering will evolve in the direction of self-improvement and enable auto-research, and, in turn, smarter models keeps harnesses simple.

Even when many harness improvement get eventually internalized into core model, the need to specify goals and context will not disappear.

[Boris Cherny @bcherny]

Proud to work at Anthropic.

[Li Yiteng @liyitengx]

Running Pi0.5 on AlohaMini (~$600 open-source BOM).

• 20-episode fine-tuning
• Action-space aligned
• Randomized object placement

Experience: aligning the action space with Pi0.5 noticeably improved the task success rate.

Exploring how far low-cost embodied hardware can go.

@LeRobotHF @lachygroom

[Tomas Ramanauskas @Tomas0413]

@eigenron I've been curating open source robotics projects — 30+ hardware builds, many with community builds showing how people actually put them together. Also has a glossary, courses, research papers if you want to go deeper. Might be a useful starting point: https://t.co/UGUazOEl53

[Jim Fan @DrJimFan]

New milestone: we trained a robot foundation model on a world model backbone, and enabled zero-shot, open-world prompting capability for new verbs, nouns, and environments. If the world model can "dream" the right future in pixels, then the robot can execute well in motors. We call it "DreamZero", our first World Action Model (WAM).

Our team had tons of fun at the lab typing anything we like into an open text prompt, and watch the robot perform tasks it was never trained on. An emergent capability we didn't quite expect. Obviously not GPT-3 reliable yet, but we are marching into the GPT-2 era.

Discoveries:

- Model and data recipe co-evolve. Compared to VLAs, WAMs learn best from diverse data, breaking away from the conventional wisdom that lots of repeated demos per task are the bread and butter. Diversity >> repetitions.

- X-embodiment is extremely hard. Pixels are the answer. Different robot morphologies traditionally have a hard time sharing knowledge well. But if we put video first, pixels become the universal bridge connecting different hardware - even videos of human first-person view.
DreamZero shows significant robot2robot and human2robot transfer. With only 55 trajectories on a *new*, unseen hardware (~30 min of teleop), it adapts so quickly and retains zero-shot prompting ability.

Yesterday I posted about the "Second Pre-training Paradigm": world models are the next-gen foundation of Physical AI, not language backbones.
Today, we are proving it works. And 2026 has just begun.

Paper: World Action Models are Zero-Shot Policies.

Read it now: (thread)

[Tomas Ramanauskas @Tomas0413]

Cool project! I added it here next to other great open source robotics projects https://t.co/UAQoCcV59O

[Li Yiteng @liyitengx]

AlohaMini 5-axis has successfully run the Pi-0.5 model. Although it still needs a small amount of human assistance (20 episodes is really too few), its performance already far exceeds that of ACT and DP under the same data scale.

[Asimov @asimovinc]

We're sharing how Asimov learned to walk. Two posts coming:

- Closing the sim2real gap (making simulation match reality)
- The walking policy powering Asimov's first steps

Sim2real tomorrow, walking policy next week.

[Tomas Ramanauskas @Tomas0413]

Amazing resource, thanks for sharing!

[Sourish Jasti @SourishJasti]

1/ General-purpose robotics is the rare technological frontier where the US / China started at roughly the same time and there's no clear winner yet.

To better understand the landscape, @zoeytang_1007, @intelchentwo, @vishnuman0 and I spent the last ~8 weeks creating a deep dive on humanoid robotics hardware and flew to China to see the supply chain firsthand.

Here's everything we've created + our takeaways about the components, humanoid comparisons, supply chains, and geopolitics👇

[Tomas Ramanauskas @Tomas0413]

Quick update: now in 8 languages. Tried to make a slightly better video, this time with sound.

https://t.co/5SeGVuDsZX

For those interested in updates: https://t.co/TY7b3MxscP

[Tomas Ramanauskas @Tomas0413]

@IlirAliu_ Here’s a short website tour video

https://t.co/eLNhwAj7Vd

[Tomas Ramanauskas @Tomas0413]

现在支持中文！https://t.co/3XgdgwzgW5

[ZeroToRobot @ZeroToRobot]

@IlirAliu_ 该资源库为机器人领域发展提供了宝贵的开源资源，值得关注和积极参与

[Carlos DP 🤖🇺🇸 @carlosdponx]

The Physical AI hackathon at finc is going great! So many teams trying stuff with robots, even a few humanoids in the mix

[Tomas Ramanauskas @Tomas0413]

¡Ahora en español! https://t.co/2ZjzXt8QiK

[Paulo Andrés Vélez Á @pvelezangel]

Robotica open source https://t.co/AwgD7QqIKZ

[Daniel Han @danielhanchen]

We use Claude Code and Codex a lot at Unsloth, and made a guide on using local models via llama.cpp and our dynamic quants - they work pretty well!

[Asimov @asimovinc]

https://t.co/sinXBaZcS5

[Ken Goldberg @Ken_Goldberg]

Looking fwd to talking about the future of Physical Intelligence at Manny's on Mission St in SF next Wed:
https://t.co/5SX6py5SMe

[Tomas Ramanauskas @Tomas0413]

Great project! I made it easy to see 3D view online:

https://t.co/RKwZZeoGqC

[Asimov @asimovinc]

We're open-sourcing Asimov Legs, a bipedal robotic system.

It's a complete leg design with 6 DOF per leg, RSU ankle architecture, passive toe joints.

Built with off-the-shelf components and compatible with MJF 3D printing.

What we're open-sourcing & sharing:
- Full mechanical CAD (STEP files)
- Actuators list
- XML files for simulation (MuJoCo)

Repo for all: https://t.co/pdCm1LYFHF

If you want to support the project, visit https://t.co/vUvQvJ97hT, for partnership DM us.

[michael avendi @avendi01]

https://t.co/TEPrbTm1pu

[KEN @cucu_eric]

@IlirAliu_ Nice

[Tuo Liu @Robo_Tuo]

Automation and robotics have massive potential to take on a lot of work in agriculture.