metadata
license: mit
pipeline_tag: summarization
ELF: Embedded Language Flows
ELF (Embedded Language Flows) is a class of continuous diffusion language models based on continuous-time Flow Matching. Unlike existing diffusion language models (DLMs), ELF predominantly stays within the continuous embedding space until the final time step, where it maps to discrete tokens using a shared-weight network. This formulation makes it straightforward to adapt established techniques from image-domain diffusion models, such as classifier-free guidance (CFG).
- Paper: ELF: Embedded Language Flows
- Repository: https://github.com/lillian039/ELF
Sample Usage
To evaluate this model on the XSum summarization task using the official JAX implementation, you can use the following command:
cd src/
# Run evaluation on XSum
python eval.py \
--config configs/training_configs/train_xsum_ELF-B.yml \
--checkpoint_path embedded-language-flows/ELF-B-xsum
The evaluator runs on the task's validation set and reports ROUGE-1/2/L scores.
Citation
If you find this work useful, please consider citing the paper:
@article{elf2026,
title={ELF: Embedded Language Flows},
author={Hu, Keya and Qiu, Linlu and Lu, Yiyang and Zhao, Hanhong and Li, Tianhong and Kim, Yoon and Andreas, Jacob and He, Kaiming},
journal={arXiv preprint arXiv:2605.10938},
year={2026}
}