jupiterelastica

With the risk of sounding naive: How come this post is flagged?

Reading the guidelines I can't see how this is off-topic or does _not_ "[gratify] one's intellectual curiosity."

Edit: spelling

I watched the Tudor Monastery Farm show first by random chance and also liked it a lot. Thanks for all the recommendations!

Additionally to locking the exposrue time and aperture, one could also take multiple exposures, figure out the camera's light response function and fuse multiple exposures together into a single higher dynamic range (HDR) image (see OpenCV tutorial on that or Debevec et al. 1997) Assuming you can find the camera response for the very long exposure times at night _and_ the very short during the day, one could relate them to each other and display both for accurate visual comparison.

Is it possible integrate this in a notebook environment? This would be amazing for teaching C/C++.

What is LORA in this context? Google only brought me to wifi networks :D

I try to. One similar thing comes to my mind: generative adversarial networks (GANs). If I'm not mistaken this is along the line of your idea of composing single ML models to bigger information processing units.

Do you, by any chance, have links or recommendations for material to read up on architectures that do consider ML models as composable gates?

Like, a generative model is a source of _some_ information that is refined with gates (classification models) conditional on the generated information?

The analogy to transistors and logic gates falls flat a bit when you consider that voltage is a rather simple univariate signal, while generated text is quite complex and multivariate. But I understand that the main point is the composability and filtering.

So, client that was "offline" can just ask a participant to get the edits from his own last one (aka pull), send out bis own offline edits (push) and resolve from there, I See.

Actually, that cleared it up, thank you :)

In Zed/CRDTs, these two seem synonymous to me since the OP's article is the first I read about CRDTs that I grasped, or so I thought ;)

Doesn't the whole process assume that whenever any edit is done (insertion/deleteion/un-/redo) that the edit _eventually_ reaches all participants?

So if a single edit is believed to be delivered to all, but actually never made it to a single participant, who resends it? And if it's not resent, then the participants state is inconsistent from now on, no?

Edit:autocorrect

I wonder what happens when an edit can not be delivered to 1+ participant(s). Does an edit need aknowledgements from _all_ participants to be persisted? Is it just dropped?

You are very clear about the current limitations on data size, which I find refreshingly honest! How sensible do you find the idea to fine tune the model to a specific problem that has more than 1000 observations, by resampling the data (similar to bootstrapping) and retraining on the subsamples? As I understand it, one could fine tune the algorithm that TabPFN learned to the specific problem.

Many thanks also for open-sourcing your work and making the colab notebook, I've been playing around with that a bit.

Edit: spelling