Instruction, demonstration, and soft-prompt adaptations produce similar accuracy but measurably different geometric reorganization of LLM representations, with evidence for cross-task synergies and trade-offs.
Format your response as follows: Joy:
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
The Geometry of Prompting: Unveiling Distinct Mechanisms of Task Adaptation in Language Models
Instruction, demonstration, and soft-prompt adaptations produce similar accuracy but measurably different geometric reorganization of LLM representations, with evidence for cross-task synergies and trade-offs.