mbeissinger

5 ms·

1. 1.▲
   Trace: Capability-Targeted Agentic Training(scalingintelligence.stanford.edu)

   1 points by mbeissinger 5mo ago|0 comments
2. 2.▲
   Meta-Harness: automated search over task-specific model harnesses(github.com)

   1 points by mbeissinger 5mo ago|0 comments
3. 3.▲
   Meta-Harness: End-to-End Optimization of Model Harnesses(yoonholee.com)

   2 points by mbeissinger 6mo ago|0 comments
4. 4.▲
   by mbeissinger 1y ago

   You might want to check out the conversational chunking from this paper: On Memory Construction and Retrieval for Personalized Conversational Agents https://arxiv.org/abs/2502.05589
5. 5.▲
   GitHub Spark: enable anyone to create software with AI and a managed runtime(githubnext.com)

   5 points by mbeissinger 2y ago|2 comments
6. 6.▲
   by mbeissinger 6y ago

   Our driving force is to make this technology as accessible as possible to as many people as possible. We believe that machine learning will be a huge new way that people interact with computers going forward to better their lives. Lobe will
7. 7.▲
   by mbeissinger 6y ago

   I don't see a way to edit the title -- is that possible for someone who isn't OP?
8. 8.▲
   by mbeissinger 6y ago

   We also want to bundle some examples with future releases so you can have data to start, great suggestion!
9. 9.▲
   by mbeissinger 6y ago

   We believe there are several advantages of Lobe over tools like Google AutoML :) Lobe is making the entire process of creating custom machine learning accessible, from creating your dataset to training and playing with your model, to integr
10. 10.▲
    by mbeissinger 6y ago

    No catch! We are first and foremost trying to make this technology accessible to as many people as possible, and we want to grow an ecosystem around it. Business models around that can come later.
11. 11.▲
    by mbeissinger 6y ago

    Yep we are starting with image classification for this initial beta launch, but plan to expand to more data types and problem types in future releases! The vision is to make a tool usable by anyone to build custom machine learning
12. 12.▲
    by mbeissinger 6y ago

    I would love to see a video of that!
13. 13.▲
    by mbeissinger 6y ago

    I believe it is this xkcd https://xkcd.com/1425/ sorry research teams :D
14. 14.▲
    by mbeissinger 6y ago

    Hey Markus from Lobe here :) all images and labels stay private to your computer, we don't ever see any of it. We only collect some generic app usage data for telemetry if you opt-in to sharing analytics after installing Lobe.
15. 15.▲
    by mbeissinger 6y ago

    Hi, Markus here from the Lobe team. The Lobe app is free and you will always be able to train for free on your computer! Our goal is to make machine learning accessible to anyone.
16. 16.▲
    by mbeissinger 8y ago

    Great work as always from hardmaru :)
17. 17.▲
    by mbeissinger 8y ago

    Thanks! Finding the right way and levels to present granularity was definitely a challenge for us developing this product.
18. 18.▲
    by mbeissinger 8y ago

    We are currently accepting applications for private beta users, and don't have public pricing information yet.
19. 19.▲
    by mbeissinger 8y ago

    Yeah! Adam and I met at NIPS 2015 when I demoed a gui prototype for OpenDeep, talking about ways to let non-coders build neural nets. Around the same time, we saw that QC Brain video Mike posted and we all started talking and unifying aroun
20. 20.▲
    by mbeissinger 8y ago

    Our vision is to be the tool that starts with great settings for beginners but lets you graduate into the internals as you become more expert - at the lowest level you can interactively create computation graphs and see their results as you
21. 21.▲
    by mbeissinger 8y ago

    Yep growing as a platform is the vision! We are committed to accessibility for people using models/components vs. paying but good point about considering the opportunity cost of someone going to algorithmia with a model once it is trai
22. 22.▲
    by mbeissinger 8y ago

    Unreal was one of the early inspirations for going the visual route - completely agree!
23. 23.▲
    by mbeissinger 8y ago

    We are focusing on free to access any model explicitly shared by the user and pay for training/deploy resources as a service, but might consider mixing in a paid route for users to monetize their unique trained models.
24. 24.▲
    by mbeissinger 8y ago

    We are also exploring exporting to other formats to run in different environments locally, or even code generation.
25. 25.▲
    by mbeissinger 8y ago

    Yes definitely, we support desktop currently with the Tensorflow SavedModel download if you need the model locally, or using the REST API for an application with online connectivity.
26. 26.▲
    by mbeissinger 8y ago

    Thanks! Acumos has a similar vision of making it easy to build, share, and deploy AI. We are also heavily focused on the iteration and feedback loop time when developing AI applications, and importantly on helping users understand what is g
27. 27.▲
    by mbeissinger 8y ago

    We plan to support when we get time series data and video upload implemented.
28. 28.▲
    by mbeissinger 8y ago

    Thanks! Yeah our approach is that diverse applications need to be able to go in and customize the models to be useful, at least for the next few years until the algorithms for AutoML get better and replace the engineers :P
29. 29.▲
    by mbeissinger 8y ago

    Web service with the option to download a compiled Tensorflow SavedModel or CoreML file if you need the model locally.
30. 30.▲
    by mbeissinger 8y ago

    We are working with private beta users now, will have pricing up when it is public.