Post-training reduces LLMs' behavioral alignment with humans across families and sizes, with the misalignment increasing in newer generations while persona induction fails to improve individual-level predictions.
The submitted datasets underwent a lightweight review process to ensure that they did not contain obvious formatting or implementation bugs
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Post-training makes large language models less human-like
Post-training reduces LLMs' behavioral alignment with humans across families and sizes, with the misalignment increasing in newer generations while persona induction fails to improve individual-level predictions.