Show HN: A Layer for robotics dataset quality

#1

Post by omertt27 » Wed, Aug 26, 2026, 1:26 PM UTC

We are exploring how teams can automatically audit data quality, detect problematic demonstrations, and build smaller, high-quality training sets before spending GPU time.

We are looking for design partners to help shape the Calibra.

Show HN: A Layer for robotics dataset quality
calibrarobotics.com

Re: Show HN: A Layer for robotics dataset quality

#2

Post by lolololomoro » Wed, Aug 26, 2026, 1:32 PM UTC

How do you measure them ?

Re: Show HN: A Layer for robotics dataset quality

#3

Post by omertt27 » Wed, Aug 26, 2026, 1:33 PM UTC

lolololomoro wrote: post #2

How do you measure them ?

you can check our repo : https://github.com/omertt27/Calibra

Re: Show HN: A Layer for robotics dataset quality

#4

Post by leroysane » Wed, Aug 26, 2026, 1:37 PM UTC

omertt27 wrote: post #3

lolololomoro wrote: post #2

How do you measure them ?

you can check our repo : https://github.com/omertt27/Calibra

[dead]

Re: Show HN: A Layer for robotics dataset quality

#5

Post by ikaru2lt » Wed, Aug 26, 2026, 3:46 PM UTC

It's really impressive to see the quality of the datasets again; in most AIs, a bad dataset means that nothing works.

Re: Show HN: A Layer for robotics dataset quality

#6

Post by ikaru2lt » Wed, Aug 26, 2026, 3:48 PM UTC

ikaru2lt wrote: post #5

It's really impressive to see the quality of the datasets again; in most AIs, a bad dataset means that nothing works.

But how do you make it better for the GPU if the size is too large?

Re: Show HN: A Layer for robotics dataset quality

#7

Post by omertt27 » Thu, Aug 27, 2026, 9:27 AM UTC

ikaru2lt wrote: post #6

ikaru2lt wrote: post #5

It's really impressive to see the quality of the datasets again; in most AIs, a bad dataset means that nothing works.

But how do you make it better for the GPU if the size is too large?

Thanks for the answer.We are approaching this with reducing unnecessary data rather than making the GPU handle more.So it analyzes quality, coverage, and redundancy to identify a smaller subset of demonstrations to use less data and compute costs.

Re: Show HN: A Layer for robotics dataset quality

#8

Post by techsage » Wed, Sep 02, 2026, 9:41 AM UTC

[dead]