Brebross
2.4K posts
I am in the @thearcitects genesis. seat #1200.
inscribed once on ethereum.
apply with your evm address ↓
ref ARC-12009C9D arcitects.fun
x.com/thearcitects/s…
Arcitects (build arc.)@TheArcitects
whitelist is open. drop wallet below arcitects.fun
English
Brebross がリツイート
Brebross がリツイート
setelah event besar top refferal leaderboard sudah selesai, akan ada event besar apalagi nih dari @axisrobotics ?
kita tunggu gebrakan di masa depan nanti👀
keep building
gAxis
Indonesia
proud to be the champion of the referral event from @axisrobotics , thanks for the reward team @plpiaoliang @chris_anm01
English
Brebross がリツイート
Axis shows that better robot data matters more than just more data. As Axis dataset size grows, performance keeps improving, especially in camera and layout robustness. The results suggest diverse tasks, scenes, and augmentations help robots generalize better, not just memorize.
Axis Robotics@axisrobotics
In our conference submission, we evaluate AXIS as a growable data engine for robot manipulation through three questions: 1. Does AXIS pretraining improve π0.5 on downstream LIBERO-Plus robustness tasks, beyond a matched-volume baseline? 2. Does the gain scale with AXIS data volume, from 25% to 50% to 100% of data volume? 3. Which perturbation axes benefit the most, and do they match the diversity targeted by our augmentation pipeline? Here, “AXIS” refers to our growable manipulation dataset snapshot built around a Franka Research 3 robot: 207 tabletop tasks across 7 scene categories, 50k+ human demonstrations, and 60k+ task/scene variants produced through cleaning and semantic-preserving augmentation. Findings below 🧵
English
Indonesia
Brebross がリツイート
NIh aku sharing buat yang penasaran gimana sih cara kontribusi atau grinding di project web3 yang fokus ke robotika atau Physical AI ✨👌
Disini aku kasi liat visual saat ngelakuin teleoperasi pake 3 project yang lagi aku serius jalanin :
🤖 @axisrobotics
🤖 @BitRobotNetwork
🤖 @ZenO4AI
Pada masing-masing project ini kita nnti akan diberi tugas yang harus diselesaikan, contohnya kyk taruh wortel dipiring, nyusun balok supaya sejajar, angkat balok dan sebagainya dari yang gampang smpe sulit 😭
Intinya kita cuma perlu nyelesaiin tugas sesuai perintahnya aja, awalnya pasti sulit aku juga ngerasain gitu 😭 tapi kalo kalian udh paham mekanismenya pasti bisa kok 👌
Tapi yang perlu kalian ketahui masing-masing project punya mekanisme dan kontrol yang berbeda jadi kalian harus bisa beradaptasi 😉
Semangat grindingnya ya kawan ✨🔥
Indonesia
@axisrobotics @AxisRoboticsID 1. CNET / AOL
🔗 aol.com/articles/2026-…
2. Business Insider (via Dnyuz)
🔗 dnyuz.com/2026/01/16/an-…
3. MIT Technology Review
🔗 technologyreview.com/2026/02/23/113…
4. SVRC / Robotics Center — Humanoid Buyer's Guide 2026 🔗 roboticscenter.ai/learn/humanoid…
English
@axisrobotics @AxisRoboticsID Even the hottest humanoid can't skip human demonstrations.
AXIS takes the same human teaching and moves it into a browser sim: no $20K hardware, no stranger watching your living room, thousands of people contributing in parallel.
Same need
Opposite approach
English
English
We are now giving out FREE inference credits! (Claude Fable included)
A) 2 ways to access free credits:
→ "@OxTrenchor can I get inference credits to build my agent?"
→ "@OxTrenchor delegate inference credits to @someone" — gift compute to a dev you believe in
B) The bigger the recipient's reach, the bigger the grant:
→ 1,000+ followers → $1,000/week
→ 500–999 → $200/week
No forms, no GitHub — reply comes with your claim link, sign in with X, start building.
English
So AXIS runs every trajectory through a 4-stage cleaning pipeline: quality filter, hesitation removal, Savitzky Golay smoothing, cubic spline resampling.
Result on 16 LIBERO benchmark tasks: motion jerk down 47%, acceleration spikes down 54%, while keeping over 95% of the data.
English
The monotonic scaling trend here is the ultimate proof of concept for AXIS. Hit Rate increasing from 66.5% to 79.4% (+12.9 point gain) with AXIS 100% shows that the benefit of this crowdsourced data engine hasn't even hit saturation yet, the flywheel actually works.
The real alpha is the RoboCasa baseline comparison though. Scoring 57.5% under the same trajectory count proves that scaling physical AI isn’t just a raw volume game it’s heavily dependent on the data recipe, semantic preserving randomization, and cleaning pipeline.
Seeing the massive absolute gains in Layout and Sensor Noise robustness on LIBERO Plus proves this model can actually generalize in unpredictable environments, rather than just overfitting on static setups. This is highly professional validation for a growable data engine, exceptional delivery on the benchmarks team.
Axis Robotics@axisrobotics
In our conference submission, we evaluate AXIS as a growable data engine for robot manipulation through three questions: 1. Does AXIS pretraining improve π0.5 on downstream LIBERO-Plus robustness tasks, beyond a matched-volume baseline? 2. Does the gain scale with AXIS data volume, from 25% to 50% to 100% of data volume? 3. Which perturbation axes benefit the most, and do they match the diversity targeted by our augmentation pipeline? Here, “AXIS” refers to our growable manipulation dataset snapshot built around a Franka Research 3 robot: 207 tabletop tasks across 7 scene categories, 50k+ human demonstrations, and 60k+ task/scene variants produced through cleaning and semantic-preserving augmentation. Findings below 🧵
English
Finally got my X-axis role from @axisrobotics
Proud to be part of this robotics project🤖
Rimuro@Hengtzyggltmpn
It's hard to believe how fast time flies while working on tasks and collecting trajectories at @axisrobotics and now I've already reached 328 trajectories. Soon it'll be 500. Keep grinding GAXIS🤖
English