| # References and License Notes |
|
|
| Time-Anchor is released under Apache-2.0. The project also records attribution |
| for Chronos-informed methods and terminology in `NOTICE` and `CITATION.cff`. |
|
|
| ## Chronos |
|
|
| The forecasting/tokenization design is informed by: |
|
|
| Ansari et al. "Chronos: Learning the Language of Time Series." Transactions on |
| Machine Learning Research, 2024. https://openreview.net/forum?id=gerNCVqqtR |
|
|
| The paper is published under CC BY 4.0. The official Chronos forecasting |
| implementation is Apache-2.0 licensed. |
|
|
| ## Generalized Attention Flow |
|
|
| The variable/time impact attribution implements: |
|
|
| Azarkhalili and Libbrecht. "Generalized Attention Flow: Feature Attribution for |
| Transformer Models via Maximum Flow." 2025. https://arxiv.org/abs/2502.15765 |
|
|
| The information tensors (AF, GF, AGF), the layered flow network of Algorithms |
| 1 and 2, and the barrier-method maximum flow formulation follow the paper; |
| token attributions are the input-node outflows of the optimal regularized flow. |
|
|
