Emotional RAG LLMs: Reading Comprehension for the Open Internet
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866918073933496320 |
|---|---|
| author | Reichman, Benjamin Avsian, Adar Talamadupula, Kartik Jawale, Toshish Heck, Larry |
| author_facet | Reichman, Benjamin Avsian, Adar Talamadupula, Kartik Jawale, Toshish Heck, Larry |
| contents | Queries to large language models (LLMs) can be divided into two parts: the instruction/question and the accompanying context. The context for retrieval-augmented generation (RAG) systems in most benchmarks comes from Wikipedia-like texts written in a neutral and factual tone. However, real-world RAG applications often retrieve internet-based text with diverse tones and linguistic styles, posing challenges for downstream tasks. This paper introduces (a) a dataset that transforms RAG-retrieved passages into emotionally inflected and sarcastic text, (b) an emotion translation model for adapting text to different tones, and (c) a prompt-based method to improve LLMs' pragmatic interpretation of retrieved text. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2408_11189 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Emotional RAG LLMs: Reading Comprehension for the Open Internet Reichman, Benjamin Avsian, Adar Talamadupula, Kartik Jawale, Toshish Heck, Larry Computation and Language Artificial Intelligence Information Retrieval Machine Learning Queries to large language models (LLMs) can be divided into two parts: the instruction/question and the accompanying context. The context for retrieval-augmented generation (RAG) systems in most benchmarks comes from Wikipedia-like texts written in a neutral and factual tone. However, real-world RAG applications often retrieve internet-based text with diverse tones and linguistic styles, posing challenges for downstream tasks. This paper introduces (a) a dataset that transforms RAG-retrieved passages into emotionally inflected and sarcastic text, (b) an emotion translation model for adapting text to different tones, and (c) a prompt-based method to improve LLMs' pragmatic interpretation of retrieved text. |
| title | Emotional RAG LLMs: Reading Comprehension for the Open Internet |
| topic | Computation and Language Artificial Intelligence Information Retrieval Machine Learning |
| url | https://arxiv.org/abs/2408.11189 |