The proposed gradient-magnitude balancing method consistently outperforms GradNorm, as it guarantees equal gradient magnitudes and considers update magnitude.

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• Loss-scale balancing and gradient-magnitude balancing are complementary and combining them achieves the best performance.claim0.848
  Ablation conclusion.
• gradient-magnitude balancingconcept0.838
  Addressing disparity in gradient magnitudes across tasks at the gradient level
• We find that the logarithm transformation also benefits existing gradient balancing methods.quote0.814
  Key finding showing the broader utility of the log transformation.
• The logarithm transformation also benefits existing gradient balancing methods.claim0.802
  Generalization of the loss transformation.
• Setting aggregated gradient scaling factor to maximum gradient norm performs best for task balancingclaim0.796
  Empirical finding on choice of αk in gradient normalization strategy
• DB-MTL achieves loss-scale balancing by performing logarithm transformation on each task loss, and rescales gradient magnitudes by normalizing all task gradients to comparable magnitudes using the maximum gradient norm.quote0.782
  Concise summary of the DB-MTL method from the abstract.
• Combining loss-scale and gradient-magnitude balancing achieves Δp = +1.15±0.16 on NYUv2.finding0.780
  Full DB-MTL ablation result.
• Task balancing requires simultaneous consideration of both loss scales and gradient magnitudesclaim0.777
  Core interpretive position of DB-MTL: complementarity of loss and gradient perspectives