IDK its been pretty solid (but it does mess up) which is where I come in. But it has helped me work with Databricks (read/writing from it) and train a model using it for some of our customers, though its NOT in prod.

Could have it be an automated monthly thing, like the who's hiring,who's wanting to be hired posts.

Yes! I think this can work!

Would Hackernews community allow for something like this or be interested in doing this or say, if I were to create this post (or perhaps the OP) every month, would that go against terms or still be allowed.

I think it can be allowed but still just want to confirm if the community really wants this

I saw an aspect of vulnerability in hackernews I hadn't seen prior which made things feel real atleast to me

I don't think an automated thing would work.

But I do think this thread is far too big to keep up with.

My plan was to post a similar but more focused thread in a month, and go from there.

I present to you https://openai.com/sam-and-jony/

Alternatively rubber tracks also are great, if you have one nearby.

Or just treadmills, I find them more gentle on my joints than concrete because it's slightly cushioned.

I myself am a DVD enthusiast (in so far as I have a copy of TDK trilogy and Raimi trilogy plus a few other classic movies/shows and songs from the 00s). There are a few shows that I enjoyed as a teen and the fact is I no longer have a way to even legally watch them in my country, so for me the ability to never lose those movies despite streaming platforms being around is the main motivator. (However I do not have a functional DVD player anymore which sucks).

So I think lets not shame people for what they do on their own time that affects none of us really.

Umm yes? The metro even if not a big deal in the states is like a small but quiet way it has changed public transport, plus moving freight, plus people over large distances, plus the bullet train that mixed luxury, speed and efficiency onto trains, all of these are quietly disruptive transformations, that I think we all take for granted.

Ahem did you mean "rightsizing" and "rapid growth"?

>We have successfully replaced thousands of complicated deep net time series based anomaly detectors at a FANG with statistical (nonparametric, semiparametric) process control ones.

They use 3 to 4 orders lower number of trained parameters and have just enough complexity that a team of 3 or four can handle several thousands of such streams.

Could you explain how ? Cause I am working on this essentially right now and it seems management is wanting to go the way of Deep NNs for our customers.

Without knowing details it's very hard to give specific recommendations. However if you follow that thread you will see folks have commented on what has worked for them.

In general I would recommend get Hyndman's (free) book on forecasting. That will definitely get you upto speed.

https://news.ycombinator.com/item?id=46058611

Wishing you the best.

If it's the case that you will ship the code over client's fence and be done with it, that is, no commitments regarding maintenance, then I will say do what the management wants. If you will continue to remain responsible for the ongoing performance of the tool then you will be better if choosing a model you understand.

MBAs do love their neural nets. As a data scientist you have to figure out what game you’re playing: is it the accuracy game or the marketing game? Back when I was a data scientist, I got far better results from “traditional” models than NN, and I was able to run off dozens of models some weeks to get a lot of exposure across the org. Combined with defensible accuracy, this was a winning combination for me. Sometimes you just have to give people what they want, and sometimes that’s cool modeling and a big compute spend rather than good results.

Without getting into specifics (just joined a new firm), we’re working with a bunch of billing data.

Management is leaning toward a deep learning forecasting approach — train a neural net to predict expected cost and then use multiple deviation scorers (including Wasserstein distance) to flag anomalies.

A simpler v1 is already live, and this newer approach isn’t my call. I’m still fairly new to anomaly detection, so for now I’m mostly trying to learn and ship within the existing direction rather than fight it.

How does next-token prediction work for time series data?

There is no single answer, because there are multiple architectures for foundation time-series models, such as T5, decoder-only models, and state-space models (SSMs).

For Chronos-2 (the current state of the art in time-series modeling), the setup is almost identical to that of LLMs because it is based on the T5 architecture. The main difference is that, in time-series models, tokens correspond to subintervals in the real-valued (ℝ) space. You can check the details here: https://arxiv.org/pdf/2510.15821