DeepSeek-V3 Technical Report

Papers with Code Trending 12/27/24, 04:03 AM Papers

Summary

DeepSeek-V3 is a parameter-efficient Mixture-of-Experts language model with 671B total parameters, achieving strong performance comparable to leading closed-source models while requiring only 2.788M H800 GPU hours for training.

We present DeepSeek-V3, a strong Mixture-of-Experts (MoE) language model with 671B total parameters with 37B activated for each token. To achieve efficient inference and cost-effective training, DeepSeek-V3 adopts Multi-head Latent Attention (MLA) and DeepSeekMoE architectures, which were thoroughly validated in DeepSeek-V2. Furthermore, DeepSeek-V3 pioneers an auxiliary-loss-free strategy for load balancing and sets a multi-token prediction training objective for stronger performance. We pre-train DeepSeek-V3 on 14.8 trillion diverse and high-quality tokens, followed by Supervised Fine-Tuning and Reinforcement Learning stages to fully harness its capabilities. Comprehensive evaluations reveal that DeepSeek-V3 outperforms other open-source models and achieves performance comparable to leading closed-source models. Despite its excellent performance, DeepSeek-V3 requires only 2.788M H800 GPU hours for its full training. In addition, its training process is remarkably stable. Throughout the entire training process, we did not experience any irrecoverable loss spikes or perform any rollbacks. The model checkpoints are available at https://github.com/deepseek-ai/DeepSeek-V3.

Original Article

View Cached Full Text

Cached at: 08/14/26, 09:31 AM

Paper page - DeepSeek-V3 Technical Report

Source: https://huggingface.co/papers/2412.19437 Published on Dec 27, 2024

Authors:

Abstract

DeepSeek-V3 is a parameter-efficient Mixture-of-Experts language model using MLA and DeepSeekMoE architectures, achieving high performance with efficient training and minimal computational cost.

We present DeepSeek-V3, a strongMixture-of-Experts (MoE)language model with 671B total parameters with 37B activated for each token. To achieve efficient inference and cost-effective training, DeepSeek-V3 adopts Multi-head Latent Attention (MLA) andDeepSeekMoEarchitectures, which were thoroughly validated in DeepSeek-V2. Furthermore, DeepSeek-V3 pioneers an auxiliary-loss-free strategy for load balancing and sets amulti-token predictiontraining objective for stronger performance. We pre-train DeepSeek-V3 on 14.8 trillion diverse and high-quality tokens, followed bySupervised Fine-TuningandReinforcement Learningstages to fully harness its capabilities. Comprehensive evaluations reveal that DeepSeek-V3 outperforms other open-source models and achieves performance comparable to leading closed-source models. Despite its excellent performance, DeepSeek-V3 requires only 2.788M H800 GPU hours for its full training. In addition, its training process is remarkably stable. Throughout the entire training process, we did not experience any irrecoverable loss spikes or perform any rollbacks. The model checkpoints are available at https://github.com/deepseek-ai/DeepSeek-V3.

View arXiv page View PDF GitHub104kauto Add to collection

Get this paper in your agent:

hf papers read 2412\.19437

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash