Metadata-Version: 2.4
Name: sdft-pytorch
Version: 0.0.2
Summary: SDFT - Pytorch
Author-email: Phil Wang <lucidrains@gmail.com>
License: MIT
Requires-Python: >=3.10
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: discrete-continuous-embed-readout
Requires-Dist: einops>=0.8.2
Requires-Dist: ema-pytorch
Requires-Dist: Jinja2
Requires-Dist: torch>=2.5
Requires-Dist: torch-einops-utils>=0.0.21
Provides-Extra: test
Requires-Dist: pytest; extra == "test"
Requires-Dist: x-transformers; extra == "test"
Dynamic: license-file

<img src="./sdft.png" width="450px"></img>

## SDFT - Pytorch (wip)

Explorations into the proposed SDFT, [Self-Distillation Enables Continual Learning](https://arxiv.org/abs/2601.19897), from Shenfeld et al. of MIT

## Citations

```bibtex
@misc{shenfeld2026selfdistillationenablescontinuallearning,
    title   = {Self-Distillation Enables Continual Learning}, 
    author  = {Idan Shenfeld and Mehul Damani and Jonas Hübotter and Pulkit Agrawal},
    year    = {2026},
    eprint  = {2601.19897},
    archivePrefix = {arXiv},
    primaryClass = {cs.LG},
    url     = {https://arxiv.org/abs/2601.19897}, 
}
```
