CAVEWOMAN: How Large Language Models Behave Under Linguistic Input and Output Compression
Two-channel evaluation shows output compression reduces costs while input compression increases costs and degrades accuracy across models and datasets.
Hugging Face · Daily Papers
·Morayo Danielle Adeyemi, Ryan A. Rossi
·
·▲ 4 upvotes
Este artigo está em destaque na seleção diária de papers do Hugging Face, curada pela comunidade de pesquisa em IA.
Autores: Morayo Danielle Adeyemi, Ryan A. Rossi, Franck Dernoncourt
- 4 upvotes da comunidade
- Temas: inference cost, output compression, input compression, task accuracy, realized cost, reference-text agreement
Resumo
Resumo original (em inglês), extraído do paper:
Two-channel evaluation shows output compression reduces costs while input compression increases costs and degrades accuracy across models and datasets.
