Welcome to Data Designer Discussions!

[johnnygreco]

👋 Welcome!

We’re using Discussions as a place to connect with other members of our community. We hope that you:

• Ask questions you’re wondering about.
• Share ideas.
• Engage with other community members.
• Welcome others and are open-minded. Remember that this is a community we
  build together 💪.

To get started, comment below with an introduction of yourself and tell us about what you do with this community.

[Pipboyguy]

Hi, I'm Marcel.

I've used this repo already create seed data for my application. It did the job amazingly, though the speed was a bit slow

[johnnygreco]

Hi @Pipboyguy 👋

Thank you for your feedback! Out of curiosity, what model provider did you try? Our research team is actively working on ways to create large datasets as fast as possible, but for sure, inference throughput is a major bottleneck.

[Pipboyguy]

Hey @johnnygreco

I first tried the models listed on data generation but they are clearly meant for robotic use cases. After some experimentation I found most 14B models like ministral-14b-instruct-2512 to be satisfactory ito speed and generated data quality/diversity.

I'm testing this with clients as well for ML applications, so its already proven economically useful! Thank you for this work.