MUX-PLMs: Pre-training Language Models with Data Multiplexing
Vishvak Murahari, Ameet Deshpande, Carlos Jimenez, Izhak Shafran, Mingqiu Wang, Yuan Cao, Karthik Narasimhan
Abstract
The widespread adoption of large language models such as ChatGPT and Bard has led to unprecedented demand for these technologies. The burgeoning cost of inference for ever-increasing model sizes coupled with hardware shortages has limited affordable access and poses a pressing need for efficiency approaches geared towards high throughput and performance. Multi-input multi-output (MIMO) algorithms such as data multiplexing, offer a promising solution with a many-fold increase in throughput by performing inference for multiple inputs at the cost of a single input. Yet these approaches are not currently performant enough to be deployed in modern systems. We change that by developing MUX-PLMs, a class of high throughput pre-trained language models (PLMs) trained with data multiplexing, that can be fine-tuned for any downstream task to yield high-throughput high-performance. Our novel multiplexing and demultiplexing modules proficiently entangle and disentangle inputs, and enable high-performance high throughput that are competitive with vanilla PLMs while achieving 2x/5x inference speedup with only a 1−4% drop on a broad suite of tasks.- Anthology ID:
- 2023.repl4nlp-1.17
- Volume:
- Proceedings of the 8th Workshop on Representation Learning for NLP (RepL4NLP 2023)
- Month:
- July
- Year:
- 2023
- Address:
- Toronto, Canada
- Editors:
- Burcu Can, Maximilian Mozes, Samuel Cahyawijaya, Naomi Saphra, Nora Kassner, Shauli Ravfogel, Abhilasha Ravichander, Chen Zhao, Isabelle Augenstein, Anna Rogers, Kyunghyun Cho, Edward Grefenstette, Lena Voita
- Venue:
- RepL4NLP
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 196–211
- Language:
- URL:
- https://aclanthology.org/2023.repl4nlp-1.17
- DOI:
- 10.18653/v1/2023.repl4nlp-1.17
- Cite (ACL):
- Vishvak Murahari, Ameet Deshpande, Carlos Jimenez, Izhak Shafran, Mingqiu Wang, Yuan Cao, and Karthik Narasimhan. 2023. MUX-PLMs: Pre-training Language Models with Data Multiplexing. In Proceedings of the 8th Workshop on Representation Learning for NLP (RepL4NLP 2023), pages 196–211, Toronto, Canada. Association for Computational Linguistics.
- Cite (Informal):
- MUX-PLMs: Pre-training Language Models with Data Multiplexing (Murahari et al., RepL4NLP 2023)
- PDF:
- https://preview.aclanthology.org/nschneid-patch-2/2023.repl4nlp-1.17.pdf