Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Lag-Llama: Towards Foundation Models for Probabilistic Time Series Forecasting
| Date | Stars |
|---|---|
| 2026-07-31 | 1598 |
| 2026-08-04 | 1597 |
| 2026-08-06 | 1597 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
# Lag-Llama: Towards Foundation Models for Probabilistic Time Series Forecasting  Lag-Llama is the <b>first open-source foundation model for time series forecasting</b>! [[Tweet Thread](https://twitter.com/arjunashok37/status/1755261111233114165)] [[Model Weights](https://huggingface.co/time-series-foundation-models/Lag-Llama)] [[Colab Demo 1: Zero-Shot Forecasting](https://colab.research.google.com/drive/1DRAzLUPxsd-0r8b-o4nlyFXrjw_ZajJJ?usp=sharing)] [[Colab Demo 2: (Preliminary Finetuning)](https://colab.research.google.com/drive/1uvTmh-pe1zO5TeaaRVDdoEWJ5dFDI-pA?usp=sharing)] [[Paper](https://arxiv.org/abs/2310.08278)] [[Video](https://www.youtube.com/watch?v=Mf2FOzDPxck)] ____ <b>Updates</b>: * **27-June-2024**: Fixed critical issues in the kv_cache implementation, improving forecast accuracy. The fixes include: resetting the self.y_cache flag globally, using causal attention correctly during kv_cache initialization, and adjusting rotary embeddings post-concatenation. Contribution by [@KelianM](https://github.com/KelianM). * **16-Apr-2024**: Released pretraining and finetuning scripts to replicate the experiments in the paper. See [Reproducing Experiments in the Paper](https://github.com/time-series-foundation-models/lag-llama?tab=readme-ov-file#reproducing-experiments-in-the-paper) for details. * **9-Apr-2024**: We have released a 15-minute video 🎥 on Lag-Llama on [YouTube](https://www.youtube.com/watch?v=Mf2FOzDPxck). * **5-Apr-2024**: Added a [section](https://colab.research.google.com/drive/1DRAzLUPxsd-0r8b-o4nlyFXrjw_ZajJJ?authuser=1#scrollTo=Mj9LXMpJ01d7&line=6&uniqifier=1) in Colab Demo 1 on the importance of tuning the context length for zero-shot forecasting. Added a [best practices section](https://github.com/time-series-foundation-models/lag-llama?tab=readme-ov-file#best-practices) in the README; added recommendations for finetuning. These recommendations will be demonstrated with an example in [Colab Demo 2](https://colab.research.google.com/drive/1uvTmh-pe1zO5TeaaRVDdoEWJ5dFDI-pA?usp=sharing) soon. * **4-Apr-2024**: We have updated our requirements file with new versions of certain packages. Please update/recreate your environments if you have previously used the code locally. * **7-Mar-2024**: We have released a preliminary [Colab Demo 2](https://colab.research.google.com/drive/1uvTmh-pe1zO5TeaaRVDdoEWJ5dFDI-pA?usp=sharing) for finetuning. Please note this is a preliminary tutorial. We recommend taking a look at the best practices if you are finetuning the model or using it for benchmarking. * **17-Feb-2024**: We have released a new updated [Colab Demo 1](https://colab.research.google.com/drive/1DRAzLUPxsd-0r8b-o4nlyFXrjw_ZajJJ?usp=sharing) for zero-shot forecasting that shows how one can load time series of different formats. * **7-Feb-2024**: We released Lag-Llama, with open-source model checkpoints and a Colab Demo for zero-shot forecasting. ____ **Current Features**: 💫 <b>Zero-shot forecasting</b> on a dataset of <b>any frequency</b> for <b>any prediction length</b>, using <a href="https://colab.research.google.com/drive/1DRAzLUPxsd-0r8b-o4nlyFXrjw_ZajJJ?usp=sharing" target="_blank">Colab Demo 1.</a><br/> 💫 <b>Finetuning</b> on a dataset using [Colab Demo 2](https://colab.research.google.com/drive/1uvTmh-pe1zO5TeaaRVDdoEWJ5dFDI-pA?usp=sharing). 💫 <b>Reproducing</b> experiments in the paper using the released scripts. See [Reproducing Experiments in the Paper](https://github.com/time-series-foundation-models/lag-llama?tab=readme-ov-file#reproducing-experiments-in-the-paper) for details. **Note**: Please see the [best practices section](https://github.com/time-series-foundation-models/lag-llama?tab=readme-ov-file#best-practices) when using the model for zero-shot prediction and finetuning. ____ ## Reproducing Experiments in the Paper To replicate the pretraining setup used in the paper, please see [the pretraining script](scripts/pretrain
Excerpt of 7,122 characters
Read on GitHub
Arjun Ashok · Canada
58
Kashif Rasul · Germany
11
1
1
Would you bet a product on this? Bounded 0–100 and slow moving.
matched fp:670e79c695ef813f, topic:foundation-models, topic:llama