Method

Negative sampling

Choosing negative examples or sampled softmax with logQ correction for retrieval training.

45 cases from 27 companies apply Negative sampling. 22 were checked in an online A/B test; 8 report results in production. Includes variants: In-batch negatives, Sampled softmax.

Typical tasks: Retrieve and rank from millions of items, Recommend from what the user just did, Rank search results, Learn user and item embeddings and 8 more. Companies: Pinterest, Yandex, LinkedIn, Criteo, Airbnb, Expedia Group and 21 more. Often combined with: Two-tower model, In-batch negatives, Sampled softmax, Approximate nearest neighbors, Multi-task learning and 78 more.

CASTLE: training natural language search before the first real querytravel2026-10
Building a Transformer-Based Category Recommender at Thumbtackmarketplace2026-09
GenRec: Towards LLM-Native Recommendation at Netflixmedia-streaming2026-07
Introducing CLEPR, our model for semantic understandingadtech2026-06
A Hybrid Generative-Ranking Model Based on Semantic IDs in Yandex Music Recommendationsmedia-streaming2026-05
X-split ARGUS: An Autoregressive Encoder-Decoder for Ad Rankingadtech2026-05
From Clicks to Conversions: Architecting Shopping Conversion Candidate Generation at Pinterestadtech2026-04
Scaling Recommendation Systems with Request-Level Deduplicationsocial2026-04
Leveraging Commerce Data for Outcome-Based Relevancy in Agentic Recommendation Systemsadtech2026-02
Ads Candidate Generation using Behavioral Sequence Modelingsocial2026-01
Contextual Property Embeddings for Corse-grained Personalizationtravel2025-07
Next-Level Personalization: How 16k+ Lifelong User Actions Supercharge Pinterest’s Recommendationssocial2025-06
Modernizing Home Feed Pre-Ranking Stagesocial2025-05
Scaling Recommender Systemsmedia-streaming2025-05
Experience with Transformer-Based Personalization at Avitoclassifieds2025-05
Scaling Loss Functions for Sequential Recommendations with Large Item Catalogsgeneral2025-05
Do We Need Graph Neural Networks for Encoding Items in Recommendations?e-commerce2025-05
Cut Cross Entropy: Efficient Training of Transformers for Recommender Systemsgeneral2025-05
Evolving Feed Ranking at Whatnotmarketplace2025-04
Embedding-based retrieval for Airbnb searchtravel2025-03
Establishing a Large Scale Learned Retrieval System at Pinterestsocial2025-01
How Instacart Uses Machine Learning to Suggest Replacements for Out-of-Stock Productsretail2024-11
Machine Learning in Content Moderation at Etsymarketplace2024-08
Candidate Generation in a Large Scale Graph Recommendation System: People You May Knowsocial2024-06
Bundled Product Recommendationsretail2024-05
Predicting a Listing Category from Its Description and Photoclassifieds2024-05
How to Build Multilingual Product Matching Without Dataretail2024-05
Geo Embeddings: Universal Representations for Locationsbanking2024-05
Decoding the Regional Fashion Signatures using AIe-commerce2024-04
Candidate Generation Using a Two Tower Approach With Expedia Group Traveler Datatravel2023-12
How LinkedIn Is Using Embeddings to Up Its Match Game for Job Seekerssocial2023-10
Embedding-based Retrieval with Two-Tower Models in Spotlightsocial2023-06
lyft2vec — Embeddings at Lyftmobility2023-03
Building and maintaining the skills taxonomy that powers LinkedIn's Skills Graphsocial2023-03
How Instacart Uses Embeddings to Improve Search Relevancee-commerce2022-09
Connecting the Dots: strengthening recommendations for our customers (Part One)food-delivery2022-06
Modern training methods for retrieval-based dialogue modelsgeneral2022-06
Matching bank and telecom customers: 3rd place at Data Fusion Contest 2022banking2022-06
Transformers for Personalizationsearch-engine2022-06
Machine Learning and Personalization in Auto.ru Searchclassifieds2022-06
Improving job matching with machine-learned activity featuresmarketplace2022-05
Beyond Matrix Factorization: Using hybrid features for user-business recommendationsmarketplace2022-04
Building a Deep Learning Based Retrieval System for Personalized Recommendationsmarketplace2022-03
Introducing Natural Language Search for Podcast Episodesmedia-streaming2022-03
Lightweight Representation Learning For Efficient And Scalable Recommendationadtech2022-01