郡主Christine (✱,✱)
@0xsexybanana
Contributor @axisrobotics
also like to share my thoughts on farming and trading.
2.5K Following 41.8K Followers
Watching Axis community training those Franka arms every day makes me think that in the future, more and more repetitive work in the physical world will be replaced by machines, and what is left for human beings to think about is:
love, longevity, and all sorts of creative work.
Show more
ray dalio spent his 77th birthday in an ibiza club until 4am
somehow this one didn't make it into Principles
kaito is getting smarter and better
Perspectives
you did it better than me!
Over 2000 bookmarks on my previous video on how to Play Axis Robotic 😳😧
So I decided to bless you all with a new 2 minutes interaction with @axisrobotics and how easy some training can be…
As of today I think I’m the best Robotic trainer, I should be hired by Elon Musk.
Show more
i have 200 k worth of #mega# during IDO,
Most of the people who participated in the IDO and decided to lock for one year were believing in the team
Now we know that:
1) mega CEO @yilongl_megaeth hasn’t lost for over a year now
2) teams’ COO Laura is left to join pharos
3) head of Product is left
4) No usdm buy back for long time
5) only token dump and mega going to zero
6) the 50m raise from IDO are used to pay team salaries since megaeth as chain has very minimal income to cover their expense
We all know what this leads to: next scroll or next bera.
And there is nothing for our one year lockers to stop loosing on this investment.
If you were me, what can do you do now?
Show more
For embodied AI to reach its GPT Moment, the data gap must first be crossed. Axis Robotics strives to build the compounding data engine for Physical AI.
We have built an end-to-end data pipeline covering simulation data, egocentric data, model pre-training, and DAgger post-training.
Axis transforms embodied AI data collection from a static, one-way and siloed workflow into dynamic, cyclic cluster-based collection and processing.
Human demonstration data feeds the policy; gaps exposed flow back to humans operators for supplementary collection and iteration, enabling continuous dynamic dataset updates.
We’ve got plenty to-dos on our time table, let’s keep building during bear
Show more
The dataset and paper based on the previous phase of community contributions are now live：
Paper Link:
Project Page:
Dataset Link:
Github Codebase:
Using Pi0.5 + AXIS-100%, we achieved 88.8 overall success on LIBERO-Plus.
For comparison:
Vanilla Pi0.5 achieved 83.9
A RoboCasa-matched simulation baseline achieved 57.5
This clearly demonstrates that diverse, in-the-wild data provides substantial gains in model performance.
And V2 is already in progress — significantly larger in scale, covering more embodiments and a wider range of atomic capabilities.
Show more
please sign your trajectories on chain asap
Remember to sign ALL your tasks before 00:00 UTC on Monday, July 27.
Something big is coming 👀
Long term builders build long term physical AI data engine on Base
Long term games with long term builders.
NEWS: TownSquare $TOWN pre-market valuation hit $280M at the highest today in @aspecta_ai pre-TGE market.
More details👇
*𝘛𝘩𝘪𝘴 𝘪𝘴 𝘯𝘰𝘵 𝘢𝘯 𝘰𝘧𝘧𝘪𝘤𝘪𝘢𝘭 𝘦𝘯𝘥𝘰𝘳𝘴𝘦𝘮𝘦𝘯𝘵 𝘰𝘧 𝘵𝘩𝘦 𝘮𝘢𝘳𝘬𝘦𝘵 𝘢𝘤𝘵𝘪𝘷𝘪𝘵𝘺 𝘰𝘯 𝘈𝘴𝘱𝘦𝘤𝘵𝘢, 𝘯𝘦𝘪𝘵𝘩𝘦𝘳 𝘪𝘴 𝘵𝘩𝘦 𝘛𝘰𝘸𝘯𝘚𝘲𝘶𝘢𝘳𝘦 𝘵𝘦𝘢𝘮 𝘣𝘦𝘩𝘪𝘯𝘥 𝘵𝘩𝘦 𝘭𝘪𝘴𝘵𝘪𝘯𝘨 𝘰𝘧 𝘵𝘩𝘦 𝘦𝘷𝘦𝘯𝘵.
Show more
Over 450 robotics training missions are now live across SN/03: TeleArms – RoboPlayground and SN/04: @axisrobotics
That's 450 opportunities to start contributing to the world's open robotics lab. No hardware required.
Who wants access before we open it up? 🦾
Show more
Every chain has DeFi. Every chain has NFTs. Only one chain has the robotics data engine accelerating real physical AI.
That chain is @base
Me: before crypto
Me: after crypto
The Train-to-Earn era in robotics is here.
Train robots.
Earn rewards.
All in under a minute.
We challenged @axisrobotics and @InvLambda on the Axis Robotics training simulator.
One of them wasn’t expecting the result… 👇
Show more
Train AI to Earn? Yes, that's the new thing we grind.
We contribute to AI through @axisrobotics.
Watch this video tutorial I did (in Filipino).
What V1 proved:
An end-to-end sim-to-real pipeline works. Engineered diversity beats raw volume (79.4 on LIBERO-Plus), and transfers directly to real physical robots.
The bottleneck:
V1 is unidirectional. Once deployed, the model can't recover from its own failures — because the data flow stops after training.
The V2 shift:
Close the loop. Human-Gated DAgger turns model errors into targeted data collection. Axis is no longer a pipeline — it's a compounding, human-in-the-loop data engine.
Show more
