Exploring DeepSeek: A Survey on Advances, Applications, Challenges and Future Directions
Swinburne University of Technology · The University of Adelaide
Abstract
The rapid advancement of large models has led to the development of increasingly sophisticated models capable of generating diverse, personalized, and high-quality content. Among these, DeepSeek has emerged as a pivotal open-source initiative, demonstrating high performance at significantly lower computation costs compared to closed-source counterparts. This survey provides a comprehensive overview of the DeepSeek family of models, including DeepSeek-V3 and DeepSeek-R1, covering their core innovations in architecture, system pipeline, algorithm, and infrastructure. We explore their practical applications across various domains, such as healthcare, finance, and education, highlighting their impact on both…
Citation impact
- FWCI
- 337.10
- Percentile
- 100%
- References
- 164
Authors
7Topics & keywords
- Data science
- Engineering ethics
- Computer science
- Engineering
- Climate action