AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Persona-Pruner: Sculpting Lightweight Models for Role-Playing

Jinsu Kim, Jihoon Tack, Noah Lee, Jongheon Jeong

ICML 2026 · Large language models › Language, safety and interpretability · approach: Faithfulness of LLM explanations · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add persona-sensitivity filtering of queries (ablation)Persona-Pruner · Alpaca-P, Llama-3.2-3B 25% sparsityLLM-judge persona score: 78.80 -> 80.19Improved
add persona-conditioned answer rewriting (ablation)Persona-Pruner · Alpaca-P, Llama-3.2-3B 25% sparsityLLM-judge persona score: 39.36 -> 80.19Improved
add both filtering and rewriting (ablation)Persona-Pruner · Alpaca-P, Llama-3.2-3B 25% sparsityLLM-judge persona score: 32.50 -> 80.19Improved
use straight-through estimator instead of Gumbel-Softmax (ablation)Persona-Pruner · Alpaca-P, Llama-3.2-3B 25% sparsityLLM-judge persona score: 75.77 -> 80.19Improved
use straight-through estimator instead of Soft-TopK (ablation)Persona-Pruner · Alpaca-P, Llama-3.2-3B 25% sparsityLLM-judge persona score: 68.55 -> 80.19Improved
use straight-through estimator instead of Soft-TopK (ablation)Persona-Pruner · PIQA, Llama-3.2-3B 25% sparsityPIQA accuracy: 0.59 -> 0.71Improved
add answer rewriting (with filtering) (ablation)Persona-Pruner · PIQA, Llama-3.2-3B 25% sparsityPIQA accuracy: 0.74 -> 0.71Hurt
add persona-specific masks instead of one global mask learned from all personas (ablation)Persona-Pruner · Alpaca-P, 10 personasLLM-judge persona scoreImproved
use persona-sensitivity filtered Alpaca-P queries versus random Alpaca queriesLlama-3.2-3B-Instruct dense · Alpaca-P vs AlpacaLLM-judge persona score: 60.00 -> 84.86 (+24.86)Improved
add recovery finetuning (ablation)Persona-Pruner · RoleBench, Llama-3.2-3B 25% sparsityLLM-judge persona score: 81.20 -> 84.26Improved
add recovery finetuning on general Alpaca (ablation)LLM-Pruner · RoleBench, Llama-3.2-3B 25% sparsityLLM-judge persona score: 48.43 -> 30.00Hurt
add multi-source data mixture (Alpaca, Open-Orca, Dolly) instead of single-source, scaling 10k to 40k (ablation)Persona-Pruner · Alpaca-P, one personaLLM-judge persona scoreImproved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy