Google just released TimesFM (a Time Series Foundation Model) - a 200M param model that can forecast time-series data it has never seen before, with no additional fine-tuning required.
Time-series forecasting is everywhere - retail, finance, healthcare, etc. And for the longest time, this was the domain of traditional statistical methods. Then deep learning models came along and did better, but it involved long training and validation cycles before you could even test them on new data.
TimesFM changes this. All we need to do is point it at a new dataset, and it gives you a solid forecast immediately - zero-shot.
The architecture is decoder-only, the same idea as GPT. Instead of words, it works with “patches” - groups of contiguous time-points treated as tokens. The model predicts the next patch from all the ones before it.
The model was pre-trained on 100 billion real-world time-points, mostly from Google Trends and Wikipedia Pageviews - which naturally capture a huge variety of patterns across domains.
On benchmarks, zero-shot TimesFM matches PatchTST and DeepAR that were explicitly trained on those datasets, and even beats GPT-3.5 on forecasting tasks despite being far smaller.
The model is open on HuggingFace and GitHub if you want to try it.