Python SDK | Not Hacker News!

Discussion (18 comments)

Showing 18 comments

11 days ago

1 reply

I do not understand how time series can be forecast without training on data from a relevant domain. Like, would these be able to predict EEG/fMRI timeseries?

armcat

10 days ago

The promise is similar to LLMs, if you pretrain on sufficiently large timeseries datasets with sufficiently large variance/characteristics, that you will be able to transfer the model to a completely different use case that exhibits somewhat similar characteristics (in latent space). But it’s always good to check what kind of data the model was trained on, eg Chronos 2.0 training data is described in Appendix A Table 6 here: https://arxiv.org/pdf/2510.15821

srean

10 days ago

1 reply

I will always advise "start simple"

https://news.ycombinator.com/item?id=46055919

anshumankmr

10 days ago

1 reply

>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.

srean

10 days ago

2 replies

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.

Wishing you the best.

clickety_clack

10 days ago

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 they want cool modeling and a big compute spend rather than good results.

anshumankmr

10 days ago

Without getting too much into specifics (just joined a new firm) its a bunch of billing data. What management is planning is a deep learning forecasting-based approach - a neural net will be trained to predict expected costs, and we use multiple scorers including Wasserstein distance to detect when actual deviates from predictions.

But the thing is the a v1 is already live, this is being chosen not by me and I am still very new to anomaly detection so I am not trying to ruffle feathers.

anshumankmr

10 days ago

1 reply

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

ChernovAndrei

9 days ago

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

kavalg

10 days ago

1 reply

It looks like this is an SaaS with an open source client only right?

OutOfHere

10 days ago

Moreover, some of the models used as listed at https://faim.it.com/models are open models developed by third-parties, and how you host and call them is up to you.

bvan

10 days ago

1 reply

Isn’t this the ultimate black box? If a forecasting system is a black box, then you have no chance of understanding why its performance might deteriorate. Once that happens it essentially becomes a digital paper-weight.

OutOfHere

10 days ago

That's not a good argument because it's like saying that an LLM is a black box, yet we use them all day every day. The two share the same operating principles.

chwzr

10 days ago

1 reply

How does it compare to tabpfn?

ChernovAndrei

9 days ago

Limix outperforms tabpfn v2: https://arxiv.org/pdf/2509.03505

smallnix

10 days ago

Before picking this I would benchmark on my existing data using e.g. https://unit8co.github.io/darts/index.html#regression-models

BobSonOfBob

10 days ago