TaskPGM optimizes a quadratic energy over task mixtures using PMI/JSD behavioral affinities, yielding mixtures that outperform naive sampling on several 7B LLM fine-tuning benchmarks.
Alvarez-Melis and N
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Learning Task Mixtures from Task Affinities: A Probabilistic Graphical Model for Supervised Fine-Tuning
TaskPGM optimizes a quadratic energy over task mixtures using PMI/JSD behavioral affinities, yielding mixtures that outperform naive sampling on several 7B LLM fine-tuning benchmarks.