Salvato in:
Dettagli Bibliografici
Autori principali: Tadimeti, Divya, Pan, Shawn, Lanka, Sameera, Zhou, Chenghui, Hasan, Sadid
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:https://arxiv.org/abs/2606.00462
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910273786347520
author Tadimeti, Divya
Pan, Shawn
Lanka, Sameera
Zhou, Chenghui
Hasan, Sadid
author_facet Tadimeti, Divya
Pan, Shawn
Lanka, Sameera
Zhou, Chenghui
Hasan, Sadid
contents Short-form text rewriting is a constrained variant of paraphrasing in which limited context and high semantic density leave little room for variation. While large language models perform well on general paraphrasing, small language models (SLMs) often struggle with semantic fidelity and hallucination robustness in short-form settings. In this work, we present an empirical study of adapting an SLM, Phi Silica, for short-form rewrite through dataset curation, prompt distillation, parameter-efficient fine-tuning, and evaluation. We curate a dataset of short presentation-style text from public slide decks and use GPT-5-chat both to generate rewrite supervision and to conduct LLM-as-a-judge evaluation. Our results show that finetuning improves semantic fidelity, reduces hallucinations, and increases preference win rate against GPT-5-chat rewrites. The findings suggest that targeted adaptation for SLMs can substantially narrow the gap to cloud models and provide practical guidance for adapting SLMs to precision-critical rewrite tasks.
format Preprint
id arxiv_https___arxiv_org_abs_2606_00462
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Short-form Text Rewriting with Phi Silica
Tadimeti, Divya
Pan, Shawn
Lanka, Sameera
Zhou, Chenghui
Hasan, Sadid
Computation and Language
Artificial Intelligence
Machine Learning
Short-form text rewriting is a constrained variant of paraphrasing in which limited context and high semantic density leave little room for variation. While large language models perform well on general paraphrasing, small language models (SLMs) often struggle with semantic fidelity and hallucination robustness in short-form settings. In this work, we present an empirical study of adapting an SLM, Phi Silica, for short-form rewrite through dataset curation, prompt distillation, parameter-efficient fine-tuning, and evaluation. We curate a dataset of short presentation-style text from public slide decks and use GPT-5-chat both to generate rewrite supervision and to conduct LLM-as-a-judge evaluation. Our results show that finetuning improves semantic fidelity, reduces hallucinations, and increases preference win rate against GPT-5-chat rewrites. The findings suggest that targeted adaptation for SLMs can substantially narrow the gap to cloud models and provide practical guidance for adapting SLMs to precision-critical rewrite tasks.
title Short-form Text Rewriting with Phi Silica
topic Computation and Language
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2606.00462