Show HN: Quickly search, build and share AI models
picsellia.com
Show HN: Quickly search, build and share AI models
1–5 of 5 posts
Re: Show HN: Quickly search, build and share AI models
#2Then you can annotate your data with our optimized interfaces with smart tools. You can also run trainings and send all your data (checkpoints, models, results, logs) to the platform, with our API, where everything will be displayed and saved so your team can access it easily.
And once you are done if your trained model is public it is automatically added to our serving engine and you can use it freely ! I know this is a bit complex product but don't hesitate to ask me if you have any questions or need any help, here is my mail : pierre-nicolas[at]picsellia[dot]com
I would really love to have feedbacks on this and it's free so please try it and tell me what you think, see you there !
Re: Show HN: Quickly search, build and share AI models
#3Hello everyone, I'm Pierre-Nicolas, CTO at Picsell.ia. What we have build is a plateform enabling you to search for public models and datasets that users can upload on the platform, clone them and use them directly in a standardized format directly on the platform. Then you can annotate your data with our optimized interfaces with smart tools. You can also run trainings and send all your data (checkpoints, models, re…
Re: Show HN: Quickly search, build and share AI models
#4Hello everyone, I'm Pierre-Nicolas, CTO at Picsell.ia. What we have build is a plateform enabling you to search for public models and datasets that users can upload on the platform, clone them and use them directly in a standardized format directly on the platform. Then you can annotate your data with our optimized interfaces with smart tools. You can also run trainings and send all your data (checkpoints, models, re…
very interesting. You can probably branch out into "crowdsourced Datarobot" by taking the top 1% of models and powering an Automl engine using that.
Re: Show HN: Quickly search, build and share AI models
#5Once you have a big marketplace, then it's a clear value for any engineer who wants easy access to quality datasets (for experimentation, transfer learning...)