Question about Chronos-2 pretraining details

[LingFengGold]

Hi Chronos-2 team,

Thank you for releasing Chronos-2 and the detailed technical report. The use of synthetic data for enabling multivariate and covariate-aware zero-shot forecasting is very helpful.

I was wondering if you could share a bit more detail about the pretraining setup, if possible:

Synthetic data

• For synthetic univariate generators (TSI, TCM, KernelSynth, etc.), roughly how many time series were generated, and what was the typical range of series lengths?
• For synthetic multivariate data, what was the approximate scale (number of samples) and the typical number of variates per series?

Training hyperparameters

• Learning rate (and whether a schedule or warmup was used)
• Batch size (and use of gradient accumulation, if any)
• Total training steps

These details would be very useful for understanding the training choices and for guiding future model and data design.

Thanks again for your work and for sharing the model.

[abdulfatir]

Hi @LingFengGold, thanks for your interest. We do not plan to share the specific training and synthetic data details.