born in 1988. PhD. His main research interests include computer architecture and electronic design automation
born in 1997. PhD. His main research interests include computer architecture, near data processing, domain specific accelerator, machine learning system and interconnection techniques
born in 1973. PhD. professor, PhD supervisor. His main research interests include generative AI, logic reasoning, and large language model for life science
born in 1986. Bachelor. Senior Member of Technical Staff. Director of Alibaba Tongyi Lab heterogeneous computing team. His main research interest includes deep learning inference
born in 1974. PhD. IEEE fellow, ACM fellow, AAAS fellow, Chair professor. His main research interests include computer architecture, VLSI, electronic design automation, machine learning
born in 1981. PhD, professor, PhD supervisor. His main research interests include electronic design automation for domain specific architecture and new devices, architecture/circuit/device codesign and automation
Deploying Transformer models under the conventional pre-train-then-fine-tune paradigm is challenging for multi-task serving, because a full model copy for each downstream task must be maintained, quickly exhausting the storage budget. Recent algorithmic advances in parameter-efficient Transformer (PET) have shown enormous potential to mitigate the storage overhead. They share the pre-trained model among tasks and only fine-tune a small portion of task-specific parameters. Unfortunately, existing serving systems neither have flexible PET task management mechanisms nor can efficiently serve queries to different tasks in batches. Therefore, we propose PetS, a unified framework for multi-task PET serving. Specifically, different PET tasks are expressed by a unified representation in the same framework, which enables flexible PET task management. Based on the unified representation, we design a specialized PET inference engine to batch different tasks’ queries together and execute them with task-agnostic shared operators and task-specific PET operators. Equipped with the PET inference engine, PetS is more scalable with respect to the number of tasks on a single GPU device. To further improve system throughput, we propose a coordinated batching strategy taking query length, PET task type as well as system load balancing together into consideration. To improve the throughput on multiple GPU instances, we also propose a PET-migration based load balancing strategy. We evaluate PetS on platforms with single GPU, including Edge/Desktop/Server GPUs. Comprehensive experiments demonstrate that PetS supports up to 26 times more concurrent tasks and improves the serving throughput by 1.53 times and 1.63 times on desktop and server GPU nodes, respectively. On multiple GPUs, our load-balancing strategy also provides up to 29% speedup.
| 科 Family | 属数 Number of genus | 种数 Number of species | 占总种数比例 Percentage of total species (%) | 属 Genus | 种数 Number of species | 占总种数比例 Percentage of total species (%) |
|---|---|---|---|---|---|---|
| 鹅膏菌科Amanitaceae | 2 | 11 | 5.26 | 鹅膏菌属 Amanita | 10 | 4.78 |
| 小菇科 Mycenaceae | 2 | 12 | 5.74 | 丝盖伞属 Inocybe | 5 | 2.39 |
| 多孔菌科 Polyporaceae | 8 | 14 | 6.70 | 蜡蘑属 Laccaria | 5 | 2.39 |
| 红菇科 Russulaceae | 3 | 23 | 11.00 | 小皮伞属 Marasmius | 6 | 2.87 |
| 小菇属 Mycena | 11 | 5.26 | ||||
| 光柄菇属 Pluteus | 5 | 2.39 | ||||
| 红菇属 Russula | 17 | 8.13 | ||||
| 栓菌属 Trametes | 5 | 2.39 |