Liang Wenfeng
Source-listed: Founder and CEO · DeepSeek
Record snapshot: Sep 3, 2026. Coverage and affiliations may be incomplete or historical.
Selected work
5- DeepSeek-R1 incentivizes reasoning in LLMs through reinforcement learningSaved credit: Contributor · 2025Explore people connected to this work →
- DeepSeek-V3 Technical ReportSaved credit: Contributor · 2024Explore people connected to this work →
- Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse AttentionSaved credit: Contributor · 2025Explore people connected to this work →
- Insights into DeepSeek-V3: Scaling Challenges and Reflections on Hardware for AI ArchitecturesSaved credit: Contributor · 2025Explore people connected to this work →
- mHC: Manifold-Constrained Hyper-ConnectionsSaved credit: Contributor · 2025Explore people connected to this work →