Pith. sign in

It uses an online, per-step rule (no pairwise gradient ops), adding negligible overhead while remaining robust to loss-scale differences

1 Pith paper cite this work. Polarity classification is still indexing.

1 Pith paper citing it

citation-role summary

baseline 1

citation-polarity summary

fields

cs.LG 1

years

2025 1

verdicts

CONDITIONAL 1

roles

baseline 1

polarities

baseline 1

representative citing papers

Graph Coloring for Multi-Task Learning

cs.LG · 2025-09-21 · conditional · novelty 5.0

SON-GOKU measures gradient conflicts between tasks, builds a conflict graph, greedily colors it, and updates one compatible task group per step, reporting gains over multi-task baselines.

citing papers explorer

Showing 1 of 1 citing paper.

  • Graph Coloring for Multi-Task Learning cs.LG · 2025-09-21 · conditional · none · ref 4

    SON-GOKU measures gradient conflicts between tasks, builds a conflict graph, greedily colors it, and updates one compatible task group per step, reporting gains over multi-task baselines.