Skip to content

๐Ÿ“‚ Others

๐Ÿ’ฌ EMNLP2026 ยท 2 paper notes

๐Ÿ“Œ Same area in other venues: ๐Ÿง  NeurIPS2026 (363) ยท ๐ŸŽž๏ธ ECCV2026 (64) ยท ๐Ÿ“ท CVPR2026 (125) ยท ๐Ÿ”ฌ ICLR2026 (121) ยท ๐Ÿ’ฌ ACL2026 (3) ยท ๐Ÿงช ICML2026 (71)

Dissecting Transformers: A CLEAR Perspective Towards Green AI

Addressing the temporal mismatch between microsecond-scale operator execution and millisecond-scale power sensors, this paper proposes CLEAR, a software-based fine-grained energy measurement pipeline that profiles individual Transformer components across 15 models, revealing that the Attention block consumes significantly more energy per FLOP than the overall model.

Train Smarter, Not Harder: Switching Signal-Guided Training in Active Learning

Addressing the conflict in active learning between computationally burdensome retraining and warm-started fine-tuning that loses plasticity and degrades calibration early on, HYBRIDAL introduces an adaptive schedule that monitors model stabilization signals (\(\Delta\alpha\) or \(\Delta\mathrm{Acc}\)) and irreversibly switches from retraining to fine-tuning, preserving non-inferior F1 while slashing up to 49% of training time and recovering substantial calibration fidelity.