3 个仓库
End-to-end frameworks for building industrial-scale recommendation systems using deep learning.
Distinct from Recommender Frameworks: Shortlist candidates are predominantly Awesome Lists; this is a functional framework implementation.
Explore 3 awesome GitHub repositories matching artificial intelligence & ml · Neural Recommendation Frameworks. Refine with filters or upvote what's useful.
这是一个神经推荐系统框架,专为构建工业级建议引擎而设计。它作为一个机器学习流水线,实现了候选检索和多阶段排序模型,根据用户行为和偏好推荐相关项目。 该框架利用双塔检索引擎将用户和项目嵌入到共享向量空间中,以实现快速相似度搜索。它结合了推荐多样性控制器和重排序机制来惩罚冗余,同时序列化用户行为模型处理时间序列动作流,以捕捉不断演变的偏好和短期意图。 该系统涵盖了完整的推荐生命周期,包括候选检索工程、排序模型优化和用户行为建模。它还包括用于管理推荐多样性以及通过准确性指标和测试评估系统性能的机制。
Provides a complete neural recommendation system framework for industrial-scale suggestion engines.
这是一个 PyTorch 推荐框架和深度学习推荐模型,旨在生成个性化的内容预测。它作为一个分布式嵌入训练器,通过神经网络架构处理密集和稀疏特征来预测用户偏好。 该项目实现了使用专用 GPU 内核的 CUDA 优化机器学习系统,以加速嵌入查找和聚合。它采用分布式方法将海量稀疏特征表分片到多个 GPU 上,从而支持大规模模型的训练。 该系统利用双塔架构进行特征交互,并支持混合并行,结合了计算集群上的数据并行和模型并行。其功能面包括跨网络节点的分布式训练、GPU 内存优化以及基于检查点的状态恢复。
Provides an end-to-end deep learning framework for building and training industrial-scale recommendation systems.
Neural collaborative filtering is a recommendation system framework that predicts user item preferences from implicit feedback by combining generalized matrix factorization and multi-layer perceptron networks through a shared final embedding layer. It captures both linear and non-linear interactions to model user preferences from historical data. The framework executes training and evaluation runs through a configuration-driven pipeline accessible via command-line interfaces, parsing hyperparameters such as learning rates, batch sizes, and latent dimensions. It optimizes implicit feedback mod
Calculates item recommendations for users using neural network architectures that combine matrix factorization and multi-layer perceptrons.