TMS: Trajectory-Mixed Supervision for Reward-Free, On-Policy SFT
- 2026-02-03
- 1
@misc{khan2026tms,
title = {TMS: Trajectory-Mixed Supervision for Reward-Free, On-Policy SFT},
author = {Rana Muhammad Shahroz Khan and Zijie Liu and Zhen Tan and C. A. Fleming and Tianlong Chen},
year = {2026},
eprint = {2602.03073},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2602.03073},
}