A vector computed as the weight difference between GRPO-trained and SFT-trained models, when added to a base model, transfers chain-of-thought reasoning improvements across benchmarks.
Deep reinforcement learning from human preferences
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
Reasoning Vectors: Transferring Chain-of-Thought Capabilities via Task Arithmetic
A vector computed as the weight difference between GRPO-trained and SFT-trained models, when added to a base model, transfers chain-of-thought reasoning improvements across benchmarks.