Learning to Share: Selective Memory for Efficient Parallel Agentic Systems
Paper • 2602.05965 • Published
How to use jfioresi/lts-decision-transformer with PEFT:
Task type is invalid.
LoRA-finetuned Qwen3-0.6B verifier for agent memory decisions (YES/NO). See https://github.com/UCF-CRCV/autogen-lts for usage information.
Base model weights are not included; load Qwen/Qwen3-0.6B and restore the adapter plus projection layers from this repo.
Please cite this work if using this model:
@inproceedings{fioresi2026learning,
title={Learning to Share: Selective Memory for Efficient Parallel Agentic Systems},
author={Fioresi, Joseph and Kulkarni, Parth Parag and Vayani, Ashmal and Wang, Song and Shah, Mubarak},
booktitle={Proceedings of the International Conference on Machine Learning (ICML)},
year={2026}
}
arXiv link: arXiv:2602.05965
from decision_transformer import DecisionTransformer
model = DecisionTransformer.from_hub("jfioresi/lts-decision-transformer")
model.eval()