PINGALA: Prosody-Aware Decoding for Sanskrit Poetry Generation

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Jagadeeshan, Manoj Balaji, Singh, Atul, Sahith, Nallani Chakravartula, Krishna, Amrith, Goyal, Pawan
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866918408368422912
author Jagadeeshan, Manoj Balaji
Singh, Atul
Sahith, Nallani Chakravartula
Krishna, Amrith
Goyal, Pawan
author_facet Jagadeeshan, Manoj Balaji
Singh, Atul
Sahith, Nallani Chakravartula
Krishna, Amrith
Goyal, Pawan
contents Poetry generation in Sanskrit typically requires the verse to be semantically coherent and adhere to strict prosodic rules. In Sanskrit prosody, every line of a verse is typically a fixed length sequence of syllables adhering to prescribed binary patterns of syllable weights. We observe that instead of treating a verse as a monolithic sequence, segmenting them as grouped-lines leads to significant improvement in semantic coherence by 10\% with comparable metrical adherence. Specifically, PINGALA, our proposed decoding approach is designed to encourage every line to have well-formed words and our token selection biases the model towards it by preferring longer tokens. Writing in Sanskrit follows phonemic orthography, hence using a phonetically aware transliteration scheme, SLP1, increased the metrical alignment by 46\% with comparable semantic similarity, for a instruction fine-tuned large language models like Phi-4. We also introduce a new approach for reference-free evaluation using cross-encoders which achieved better alignment with true poetry instances.
format Preprint
id arxiv_https___arxiv_org_abs_2603_24413
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle PINGALA: Prosody-Aware Decoding for Sanskrit Poetry Generation
Jagadeeshan, Manoj Balaji
Singh, Atul
Sahith, Nallani Chakravartula
Krishna, Amrith
Goyal, Pawan
Computation and Language
Poetry generation in Sanskrit typically requires the verse to be semantically coherent and adhere to strict prosodic rules. In Sanskrit prosody, every line of a verse is typically a fixed length sequence of syllables adhering to prescribed binary patterns of syllable weights. We observe that instead of treating a verse as a monolithic sequence, segmenting them as grouped-lines leads to significant improvement in semantic coherence by 10\% with comparable metrical adherence. Specifically, PINGALA, our proposed decoding approach is designed to encourage every line to have well-formed words and our token selection biases the model towards it by preferring longer tokens. Writing in Sanskrit follows phonemic orthography, hence using a phonetically aware transliteration scheme, SLP1, increased the metrical alignment by 46\% with comparable semantic similarity, for a instruction fine-tuned large language models like Phi-4. We also introduce a new approach for reference-free evaluation using cross-encoders which achieved better alignment with true poetry instances.
title PINGALA: Prosody-Aware Decoding for Sanskrit Poetry Generation
topic Computation and Language
url https://arxiv.org/abs/2603.24413