DeepSeek-V3 DeepSeek open source hybrid expert model

DeepSeek-V3 is an open-source large language model built on Mixture of Experts (MoE) and Multi-Head Latent Attention (MLA) architectures, totaling 671B parameters with only 37B activated per token to greatly reduce training and inference costs while excelling at complex tasks like mathematical reasoning and code generation . Released under the MIT license and implemented in Python, DeepSeek-V3 is actively maintained by a community of 21 contributors.

Official website: https://www.deepseek.com/
Source code: https://github.com/deepseek-ai/DeepSeek-V3

Architecture: Innovative Load Balancing Strategy and Training Objective

• On top of the efficient architecture of DeepSeek-V2, we pioneer an auxiliary-loss-free strategy for load balancing, which minimizes the performance degradation that arises from encouraging load balancing.
• We investigate a Multi-Token Prediction (MTP) objective and prove it beneficial to model performance. It can also be used for speculative decoding for inference acceleration.

Pre-Training: Towards Ultimate Training Efficiency

• We design an FP8 mixed precision training framework and, for the first time, validate the feasibility and effectiveness of FP8 training on an extremely large-scale model.
• Through co-design of algorithms, frameworks, and hardware, we overcome the communication bottleneck in cross-node MoE training, nearly achieving full computation-communication overlap.
  This significantly enhances our training efficiency and reduces the training costs, enabling us to further scale up the model size without additional overhead.
• At an economical cost of only 2.664M H800 GPU hours, we complete the pre-training of DeepSeek-V3 on 14.8T tokens, producing the currently strongest open-source base model. The subsequent training stages after pre-training require only 0.1M GPU hours.

Post-Training: Knowledge Distillation from DeepSeek-R1

• We introduce an innovative methodology to distill reasoning capabilities from the long-Chain-of-Thought (CoT) model, specifically from one of the DeepSeek R1 series models, into standard LLMs, particularly DeepSeek-V3. Our pipeline elegantly incorporates the verification and reflection patterns of R1 into DeepSeek-V3 and notably improves its reasoning performance. Meanwhile, we also maintain a control over the output style and length of DeepSeek-V3.

To ensure optimal performance and flexibility, we have partnered with open-source communities and hardware vendors to provide multiple ways to run the model locally. For step-by-step guidance, check out Section 6: How_to Run_Locally.

For developers looking to dive deeper, we recommend exploring README_WEIGHTS.md for details on the Main Model weights and the Multi-Token Prediction (MTP) Modules. Please note that MTP support is currently under active development within the community, and we welcome your contributions and feedback.

Evaluation results on the Needle In A Haystack (NIAH) tests. DeepSeek-V3 performs well across all context window lengths up to 128K.