Datadog Releases Toto 2.0 Open-Weight Time-Series Models Up to 2.5 Billion Parameters
Datadog released Toto 2.0, a family of five open-weight forecasting models for time-series data ranging from 4 million to 2.5 billion parameters. The company said the weights are available on Hugging Face under an Apache 2.0 license.
Datadog said each larger model in the family outperformed the next smaller one using a single hyperparameter configuration, and that Toto 2.0 ranked first on the BOOM, GIFT-Eval and TIME benchmarks. The company said the results suggest a more predictable relationship between scale and downstream performance in time-series foundation models.
From the sources (4 posts)
@tim_dettmersRT @atalwalkar: Today we’re releasing Toto 2.0: a family of open-weights time series foundation models spanning 4M to 2.5B parameters. The…
@clementdelangueRT @datadoghq: Toto 2.0 is here: Datadog AI's 5 open-weights forecasting models (4m-2.5B params) finally make scaling work for time series…
@clementdelangueAre scaling laws finally working for time series foundation models? Today, @datadoghq is releasing Toto 2.0 weights in Apache 2.0 on @huggingface. It's a family of open-weights TSFMs from 4M to 2.5B parameters, where every size beats the l
@huggingfaceRT @ClementDelangue: Are scaling laws finally working for time series foundation models? Today, @datadoghq is releasing Toto 2.0 weights i…