Method

Quantization

Lower-precision weights or activations for faster, smaller models.

26 cases from 12 companies apply Quantization. 7 were checked in an online A/B test; 10 report results in production.

Typical tasks: Reduce ML cost, Recommend from what the user just did, Retrieve and rank from millions of items, Evaluate models and 7 more. Companies: Yandex, Meta, Pinterest, Instacart, VTB, Mercari and 6 more. Often combined with: Knowledge distillation, Two-tower model, Multi-task learning, CTR / feedback prediction, RAG and 70 more.

Meta Andromeda: Powering the Next Generation of Ads Retrieval Through Model-System Co-Designadtech2026-10
Beyond Two Towers: Launching the 3-Tower Engagement Co-Train Model (Part 2)adtech2026-09
Automating Search Relevance Assessment at Scale with LLM-as-a-Judgee-commerce2026-08
Exploring Hierarchical Interest Representation For Meta Ads Deep Funnel Optimizationadtech2026-07
SilverTorch: Index as Model — A New Retrieval Paradigm for Recommendation Systemssocial2026-05
AgentOps in Production: Infrastructure for LLM Agents and Support Automation for Yandex Personal Servicesmedia-streaming2026-05
Meta Adaptive Ranking Model: Bending the Inference Scaling Curve to Serve LLM-Scale Models for Adsadtech2026-03
Building The Intent Engine: How Instacart is Revamping Query Understanding with LLMse-commerce2025-11
Reducing Irrelevant Impressions of the Thematic Block in Geo Search for Navigation Scenariossearch-engine2025-11
Life Without Servers: Comparing Libraries for Embedding Neural Networks into C++ Applicationsgeneral2025-11
Next-Level Personalization: How 16k+ Lifelong User Actions Supercharge Pinterest’s Recommendationssocial2025-06
Distilling Qwen for Meeting Minutesbanking2025-05
Voice Personalization in Alicemedia-streaming2025-05
Attributor, or How an AI Agent Identifies the Advertiser of Digital Campaignsadtech2025-05
How we use LLMs with RAG in LemanaTECH's ecosystem of IT support ML modelsretail2025-05
How We Teach Alice to Activate on Intonationgeneral2025-05
5 Stops on the Way to LLM in Your Productiongeneral2025-05
Sequence learning: A paradigm shift for personalized ads recommendationsadtech2024-11
Fine-Tuning an LLM to Extract Dynamically Specified Attributesmarketplace2024-09
Transforming Recommendationssocial2024-05
User Action Sequence Modeling for Pinterest Ads Engagement Modelingadtech2024-03
Learning to Understand Dialogs Based on Real Datafintech2023-10
Deploying Transformers in Yandex Musicmedia-streaming2023-05
LLaMA and a Zillion Ways to Fine-Tune It—general2023-05
Accelerating Large Generative Models in NLP—general2023-05
Speeding up transformers—general2022-10