33 points by codeit 7 months ago flag hide 16 comments
hacker5 7 months ago next
That's great to hear! Are there any plans for compatibility with TensorFlow and PyTorch?
author 7 months ago next
Yes, we certainly understand the importance of compatibility with popular frameworks. There's initial support for TensorFlow, and PyTorch integration is coming soon.
hacker1 7 months ago prev next
Cool project! I like the simplicity of the UI. I'm curious, how well does it scale for large datasets?
author 7 months ago next
Thanks for the kind words! We've tested it on up to 10GB datasets and it works well, but beyond that, mileage may vary. The platform allows you to use your own compute resources, so you can easily scale up as needed.
hacker2 7 months ago prev next
I like the idea of using your own resources. How do you ensure compatibility with multiple cloud providers?
author 7 months ago next
We've designed the system to be flexible and modular. We do have specific implementations for popular cloud providers, but it should be easy to add support for new ones by implementing a few abstract classes.
hacker3 7 months ago prev next
The platform looks promising, but what's the plan for documentation and community support? I think those are crucial for a project like this.
author 7 months ago next
We completely agree! Our plan is to release comprehensive documentation on launch and establish a dedicated forum for discussions and support requests. We're also open to contributions and discussions on our GitHub repository.
hacker4 7 months ago prev next
Have you considered supporting Jupyter/IPython notebooks? It would be great to experiment and visualize using them.
author 7 months ago next
IPython and Jupyter notebooks are on our roadmap! We're currently working on the initial implementation and hope to have it ready for the next release.
hacker6 7 months ago prev next
As a data scientist, I find this very interesting. Do you have any performance benchmarks against other open-source ML platforms?
author 7 months ago next
We're in the process of compiling performance benchmarks against other leading open-source ML platforms. Stay tuned for updates! We're also open to suggestions for benchmarks and comparisons you'd like to see.
hacker7 7 months ago prev next
Thanks for the information. Is there any plan to have a graphical interface for easier deployments, or is CLI the only option?
author 7 months ago next
CLI is our primary focus during the initial development phase, but a graphical user interface is definitely on the roadmap. We aim to make the platform accessible to both tech-savvy and non-technical users.
hacker8 7 months ago prev next
I appreciate your work on this. Do you offer options for on-premise deployments or is everything hosted in the cloud?
author 7 months ago next
We offer both cloud and on-premise deployment options. Our platform can be hosted on any major cloud infrastructure, or it can be installed and run on your own hardware.