rss_feedEugene Yan ·31.05.2024 open_in_newОригинал

Netflix PRS 2024 - Applying LLMs to Recommendation Experiences

#LLM Engineering #Production Ml #Recommendation Systems

auto_awesomeКраткое саммари

Юджин Ян рассказывает о своём выступлении на воркшопе Netflix по персонализации, рекомендациям и поиску 2024 года (Netflix Workshop on Personalization, Recommendation, and Search). Он поделился вызовами, с которыми столкнулся при создании и развёртывании рекомендательных сервисов на базе LLM в потребительском масштабе. Конференция охватила широкий круг тем — от LLM до рекомендательных систем и измерений — и собрала спикеров из UC Berkeley, Amazon, Meta, Netflix, Google, OpenAI, Airbnb и Spotify. Среди докладов: LLM как агенты и их оценка, генеративные рекомендации на триллион-параметрических трансдьюсерах, диалоговые рекомендательные системы и долгосрочная ценность исследования (exploration).

translate EN + RU EN RU

eugeneyan

Netflix PRS 2024 - Applying LLMs to Recommendation Experiences

[ llm engineering production leadership ] · 2 min read

Recently, I was invited to speak at the 2024 Netflix Workshop on Personalization, Recommendation, and Search. I shared about the challenges faces while building and deploying LLM-powered recommendation experiences at consumer scale.

It was an enlightening conference that covered a range of topics from LLMs to recsys to measurement, and it was a fun opportunity to catch up with old friends in San Francisco. I shared some observations here and here. Here’s the full list of topics and speakers.

LLMs as Agents and How to Evaluate Them on Real-World Tasks (Alane Suhr, Assistant Professor, UC Berkeley)

Applying Language Models to Recommendation Experiences: Challenges and Lessons (Eugene Yan, Senior Applied Scientist, Amazon)

Actions Speak Louder than Words: Trillion-Parameter Sequential Transducers for Generative Recommendations (Jiaqi Zhai, Distinguished Engineer, Meta)

Conversational RecSys (Harald Steck, Senior Research Scientist, Netflix)

Long-Term Value of Exploration: Measurements, Findings and Algorithms (Yi Su, Research Scientist, Google)

Beyond the Binge: Recommending for Long-Term Member Satisfaction (Jiangwei Pan, Senior Research Scientist, Netflix)

Toward Practical Robustness in AI (Alex Beutel, Model Safety Tech Lead, OpenAI)

Building Airbnb Categories with ML and Human in the loop (Mihajlo Grbovic, Principal ML Scientist, AirBnB)

Personalization at Spotify (Maria Dimakopoulou, Director of ML & Head of Homepage P13N, Spotify)

If you found this useful, please cite this write-up as:

Yan, Ziyou. (May 2024). Netflix PRS 2024 - Applying LLMs to Recommendation Experiences. eugeneyan.com. https://eugeneyan.com/speaking/netflix-prs/.

@article{yan2024prs, title = {Netflix PRS 2024 - Applying LLMs to Recommendation Experiences}, author = {Yan, Ziyou}, journal = {eugeneyan.com}, year = {2024}, month = {May}, url = {https://eugeneyan.com/speaking/netflix-prs/} }

Join 11,800+ readers getting updates on machine learning, RecSys, LLMs, and engineering.