Showing 1–2 of 2 results for author: Manepalli, S

Zero-shot Conversational Summarization Evaluations with small Large Language Models

Authors: Ramesh Manuvinakurike, Saurav Sahay, Sangeeta Manepalli, Lama Nachman

Abstract: Large Language Models (LLMs) exhibit powerful summarization abilities. However, their capabilities on conversational summarization remains under explored. In this work we evaluate LLMs (approx. 10 billion parameters) on conversational summarization and showcase their performance on various prompts. We show that the summaries generated by models depend on the instructions and the performance of LLM… ▽ More Large Language Models (LLMs) exhibit powerful summarization abilities. However, their capabilities on conversational summarization remains under explored. In this work we evaluate LLMs (approx. 10 billion parameters) on conversational summarization and showcase their performance on various prompts. We show that the summaries generated by models depend on the instructions and the performance of LLMs vary with different instructions sometimes resulting steep drop in ROUGE scores if prompts are not selected carefully. We also evaluate the models with human evaluations and discuss the limitations of the models on conversational summarization △ Less

Submitted 29 November, 2023; originally announced November 2023.

Comments: Accepted at RoF0Mo workshop at Neurips 2023

D2D: Digital Archive to MPEG-21 DIDL

Authors: Suchitra Manepalli, Giridhar Manepalli, Michael L. Nelson

Abstract: Digital Archive to MPEG-21 DIDL (D2D) analyzes the contents of the digital archive and produces an MPEG-21 Digital Item Declaration Language (DIDL) encapsulating the analysis results. DIDL is an extensible XML-based language that aggregates resources and the metadata. We provide a brief report on several analysis techniques applied on the digital archive by the D2D and provide an evaluation of i… ▽ More Digital Archive to MPEG-21 DIDL (D2D) analyzes the contents of the digital archive and produces an MPEG-21 Digital Item Declaration Language (DIDL) encapsulating the analysis results. DIDL is an extensible XML-based language that aggregates resources and the metadata. We provide a brief report on several analysis techniques applied on the digital archive by the D2D and provide an evaluation of its run-time performance. △ Less

Submitted 15 February, 2006; originally announced February 2006.

Comments: 28 pages, 8 figures