Axis Robotics India

[Axis Robotics India @axisroboticsin_]

@axisrobotics Weekly

Pichhle hafte humne production task generation ko unblock karne, scene variants ko critical path se hatane, aur HG-DAgger ko ek local proof of concept se ek measurable training recipe mein convert karne par focus kiya.

Is hafte infrastructure par bhi kaam hua faster scene delivery, cleaner verify flow aur faster policy evaluation loop ke saath aur pehli baar old aur new task distributions dono par clear distillation gains dekhne ko mile.

Key updates:

Scene & Collection UX: Scene variants ab faster load hote hain, lighting rendering wapas aa gayi hai, aur post-task scoring ab zyada precise hai.

Task Generation & Throughput: LIBERO Pro, RoboCasa aur wheeled embodiments (wheeled Franka samet) ab regular generation mein hain, aur publish queue consistently 200+ tasks par bani hui hai. Policy training aur evaluation throughput lagbhag 3x ho gaya hai.

HG-DAgger Distillation: Curated-correction recipe ne old aur new dono distributions par success rate ko 20+ points improve kiya hai. (Abhi aur seeds ki zarurat hai.👇

[Axis Robotics @axisrobotics]

Axis Weekly

Last week, we focused on unblocking production task generation, moving scene variants off the critical path, and turning HG-DAgger from a local proof of concept into a measurable training recipe.

The week combined infrastructure work — faster scene delivery, a cleaner verify flow, and a quicker policy evaluation loop — with the first clear distillation gains on both old and new task distributions.

Key updates:

- Scene & collection UX: Scene variants load faster, lighting rendering is back, and post-task scoring is tighter.

- Task generation & throughput: LIBERO Pro, RoboCasa, and wheeled embodiments (incl. wheeled Franka) are now in regular generation, with the publish queue consistently above 200 tasks. Policy training and evaluation throughput roughly tripled.

- HG-DAgger distillation: A curated-correction recipe lifted success rate by 20+ points on both old and new distributions (more seeds still needed).

Details below 👇

[Axis Robotics India @axisroboticsin_]

@axisrobotics Weekly

Pichhle hafte humne production task generation ko unblock karne, scene variants ko critical path se hatane, aur HG-DAgger ko ek local proof of concept se ek measurable training recipe mein convert karne par focus kiya.

Is hafte infrastructure par bhi kaam hua faster scene delivery, cleaner verify flow aur faster policy evaluation loop ke saath aur pehli baar old aur new task distributions dono par clear distillation gains dekhne ko mile.

Key updates:

Scene & Collection UX: Scene variants ab faster load hote hain, lighting rendering wapas aa gayi hai, aur post-task scoring ab zyada precise hai.

Task Generation & Throughput: LIBERO Pro, RoboCasa aur wheeled embodiments (wheeled Franka samet) ab regular generation mein hain, aur publish queue consistently 200+ tasks par bani hui hai. Policy training aur evaluation throughput lagbhag 3x ho gaya hai.

HG-DAgger Distillation: Curated-correction recipe ne old aur new dono distributions par success rate ko 20+ points improve kiya hai. (Abhi aur seeds ki zarurat hai.👇

[Axis Robotics India @axisroboticsin_]

@axisrobotics Weekly

Pichhle hafte humne production task generation ko unblock karne, scene variants ko critical path se hatane, aur HG-DAgger ko ek local proof of concept se ek measurable training recipe mein convert karne par focus kiya.

Is hafte infrastructure par bhi kaam hua faster scene delivery, cleaner verify flow aur faster policy evaluation loop ke saath aur pehli baar old aur new task distributions dono par clear distillation gains dekhne ko mile.

Key updates:

Scene & Collection UX: Scene variants ab faster load hote hain, lighting rendering wapas aa gayi hai, aur post-task scoring ab zyada precise hai.

Task Generation & Throughput: LIBERO Pro, RoboCasa aur wheeled embodiments (wheeled Franka samet) ab regular generation mein hain, aur publish queue consistently 200+ tasks par bani hui hai. Policy training aur evaluation throughput lagbhag 3x ho gaya hai.

HG-DAgger Distillation: Curated-correction recipe ne old aur new dono distributions par success rate ko 20+ points improve kiya hai. (Abhi aur seeds ki zarurat hai.👇

[Axis Robotics @axisrobotics]

Axis Weekly

Last week, we focused on unblocking production task generation, moving scene variants off the critical path, and turning HG-DAgger from a local proof of concept into a measurable training recipe.

The week combined infrastructure work — faster scene delivery, a cleaner verify flow, and a quicker policy evaluation loop — with the first clear distillation gains on both old and new task distributions.

Key updates:

- Scene & collection UX: Scene variants load faster, lighting rendering is back, and post-task scoring is tighter.

- Task generation & throughput: LIBERO Pro, RoboCasa, and wheeled embodiments (incl. wheeled Franka) are now in regular generation, with the publish queue consistently above 200 tasks. Policy training and evaluation throughput roughly tripled.

- HG-DAgger distillation: A curated-correction recipe lifted success rate by 20+ points on both old and new distributions (more seeds still needed).

Details below 👇

[Axis Robotics India @axisroboticsin_]

Curated Corrections par DAgger Distillation

Humne model size, policy type, learning rate, training length, data mix, action chunking, clipping, MoE, LoRA aur noise ratio ke across ek broad testing ki.

Ek run mein, jisme noticeable improvement mila, humne forgetting ko limit karne ke liye curated correction data ko older distribution se distillation ke saath combine kiya older set se 200 episodes, ek focus task se 15 successful correction episodes, 24 human interventions, aur 192 correction transitions. Student model ko new data par behavior cloning ke saath train kiya gaya, aur ek frozen older model ke against matching loss bhi use kiya gaya.

Success rate old aur new dono distributions par 20+ percentage points improve hua, halanki is result ko confirm karne ke liye abhi aur seeds ki zarurat hai. Data normalizer ko freeze karne ya data ko mix karne se ek pehle degraded task par capability drop bhi kam hua.

Cleaner post tasks release kiye gaye hain, taaki isi recipe ko aur clean tareeke se dobara test kiya ja sake.

Is par ek detailed blog jald aa raha hai 👀

[Axis Robotics India @axisroboticsin_]

@axisrobotics
🔥 Indian Regional AMA Successfully Hosted🇮🇳

Aaj Indian Regional mein ek amazing Regional AMA host kiya, jise our Regional Leader
@JEAMSETH07
ne conduct kiya. 💪🔥

Event ko host karne mein
@mdkhalid768
aur
@drag4_t
ne bhi bahut achhi help ki. 🙌

AMA ke dauran humne users ko guide kiya:
📱 Mobile se tasks kaise complete karne hain

❓ Tasks ko lekar users ke questions aur issues solve kiye

Roles ke baare mein proper guidance di

Kuch important points aur useful information bhi share ki

Aur sabse badi baat 50+ members ne event mein join karke actively participate kiya🔥

Indian Regional community ki energy dekhkar genuinely kaafi amazing laga. ❤️‍🔥

Big thanks to everyone who joined and made this AMA successful🇮🇳🤝

[Axis Robotics India @axisroboticsin_]

@axisrobotics
🔥 Indian Regional AMA Successfully Hosted🇮🇳

Aaj Indian Regional mein ek amazing Regional AMA host kiya, jise our Regional Leader
@JEAMSETH07
ne conduct kiya. 💪🔥

Event ko host karne mein
@mdkhalid768
aur
@drag4_t
ne bhi bahut achhi help ki. 🙌

AMA ke dauran humne users ko guide kiya:
📱 Mobile se tasks kaise complete karne hain

❓ Tasks ko lekar users ke questions aur issues solve kiye

Roles ke baare mein proper guidance di

Kuch important points aur useful information bhi share ki

Aur sabse badi baat 50+ members ne event mein join karke actively participate kiya🔥

Indian Regional community ki energy dekhkar genuinely kaafi amazing laga. ❤️‍🔥

Big thanks to everyone who joined and made this AMA successful🇮🇳🤝

[Axis Robotics India @axisroboticsin_]

Tumhare liye kuch bada aane wala hai…

Notifications on rakho

[Axis Robotics India @axisroboticsin_]

Tumhare liye kuch bada aane wala hai…

Notifications on rakho

[Axis Robotics @axisrobotics]

Something big is coming your way…

Notifications on.

[Axis Robotics India @axisroboticsin_]

@openroboto ke saath hamari partnership announce karte hue khushi ho rahi hai, kyunki Axis Robotics apni data capabilities ko @opentensor ecosystem tak expand kar raha hai.
OpenRoboto, Bittensor (Subnet 80) par ek open competition hai jo robotics models ko continuously improve karne par focus karta hai. Miners π0.5 se shuru hone wale ek open base model ko fine-tune karte hain. Har submission ko randomized LIBERO-Pro environments mein score kiya jata hai, aur sirf wahi model naya base banta hai jo existing model se strictly better perform karta hai.
OpenRoboto ke core data engine ke roop mein, Axis 3 million se zyada multi-modal trajectories OpenRoboto ke Open Data Pool mein provide kar raha hai. Isse Bittensor network par miners ke liye physical AI model training ko accelerate karne mein madad milegi.
Iske alawa, Axis apne Data-to-Model Pipeline ke through OpenRoboto ke authoritative benchmarking platform ko support karega, jisse multiple benchmarks par data aur models ki robust aur auditable evaluation possible hogi.
Yeh partnership distributed data collection aur verifiable model evaluation ko ek strong end-to-end framework mein connect karti hai.
OpenRoboto ke baare mein aur jaanne ke liye 👇
https://t.co/vIEhetJ6vy

[Axis Robotics India @axisroboticsin_]

@openroboto ke saath hamari partnership announce karte hue khushi ho rahi hai, kyunki Axis Robotics apni data capabilities ko @opentensor ecosystem tak expand kar raha hai.
OpenRoboto, Bittensor (Subnet 80) par ek open competition hai jo robotics models ko continuously improve karne par focus karta hai. Miners π0.5 se shuru hone wale ek open base model ko fine-tune karte hain. Har submission ko randomized LIBERO-Pro environments mein score kiya jata hai, aur sirf wahi model naya base banta hai jo existing model se strictly better perform karta hai.
OpenRoboto ke core data engine ke roop mein, Axis 3 million se zyada multi-modal trajectories OpenRoboto ke Open Data Pool mein provide kar raha hai. Isse Bittensor network par miners ke liye physical AI model training ko accelerate karne mein madad milegi.
Iske alawa, Axis apne Data-to-Model Pipeline ke through OpenRoboto ke authoritative benchmarking platform ko support karega, jisse multiple benchmarks par data aur models ki robust aur auditable evaluation possible hogi.
Yeh partnership distributed data collection aur verifiable model evaluation ko ek strong end-to-end framework mein connect karti hai.
OpenRoboto ke baare mein aur jaanne ke liye 👇
https://t.co/vIEhetJ6vy

[Axis Robotics @axisrobotics]

Announcing our partnership with @OpenRoboto as Axis Robotics expands its data capabilities into the @opentensor ecosystem

OpenRoboto is an open competition on Bittensor (Subnet 80) for continuously improving robotics models. Miners fine-tune an open base model starting from π0.5, every submission is scored in randomized LIBERO-Pro environments, and only a strictly better model becomes the new base.

As the core data engine behind OpenRoboto, Axis is supplying over 3 million multi-modal trajectories into OpenRoboto’s Open Data Pool to accelerate physical AI model training for miners across the Bittensor network. Furthermore, Axis will support OpenRoboto's authoritative benchmarking platform with its Data-to-Model Pipeline, enabling robust, auditable evaluation for both data and models across multiple benchmarks.

This partnership establishes a robust, end-to-end framework uniting distributed data collection with verifiable model evaluation.

Learn more about OpenRoboto: https://t.co/9viK2AXyfy

[Axis Robotics India @axisroboticsin_]

Hum jaante hain ki gas fees kaafi zyada ho sakti hain... isliye har Saturday hum aapko rewards de rahe hain!
Axis Crazy Saturday mein welcome! Har Saturday on-chain signatures complete karein aur automatically 24-hour raffle mein enter ho jaayein.
✱ Prize: 5 winners × 20 USDT (har raffle mein total 100 USDT)
✱ Time Window: Har Saturday, 00:00–24:00 SGT → Friday 21:30 se Saturday 21:30 IST
✱ Jitne zyada tasks sign karoge = utne zyada tickets
Winners aur payouts har Monday Discord mein announce kiye jayenge.
Milte hain wahan

[Axis Robotics @axisrobotics]

We know gas fees can add up... so every Saturday, we're giving back!

Welcome to Axis Crazy Saturday! Complete on-chain signatures every Saturday to automatically enter our 24-hour raffle.

✱ Prize: 5 winners × 20 USDT (100 USDT total per raffle)
✱ Time Window: Every Saturday, 00:00–24:00 SGT
✱ More tasks signed = more tickets

Winners & payouts drop every Monday in our Discord. See you there!

[Axis Robotics India @axisroboticsin_]

Physical AI agla bada frontier hai, aur numbers khud iska momentum dikha rahe hain.
@base aur @axisrobotics ke saath 3M trajectories ka milestone achieve ho chuka hai ab target 100M ka hai.
India mein Physical AI ka ecosystem abhi bas shuru hua hai. 🇮🇳

[Axis Robotics India @axisroboticsin_]

Physical AI agla bada frontier hai, aur numbers khud iska momentum dikha rahe hain.
@base aur @axisrobotics ke saath 3M trajectories ka milestone achieve ho chuka hai ab target 100M ka hai.
India mein Physical AI ka ecosystem abhi bas shuru hua hai. 🇮🇳

[Base India @BasedIndia]

Physical AI is the next frontier.

@base is the hub for Physical AI data.

And @axisrobotics is leading the charge.

3M trajectories ✅
100M next.