korrents

korrents · piece

Training an LLM-RecSys Hybrid for Steerable Recs with Semantic IDs

Eugene Yan · 14 Sept 2025 · eugeneyan.com ↗

2 korrents from this piece

Eugene Yan did not write this page.

Every claim below was made in this piece, quoted word for word and numbered in the order the piece makes them, so you can read it there rather than take our word for it. The sentence above each quote is our reading of the claim, not their wording. Each quote was checked against a stored copy of the page at build time; where the two differ, the quote is the fact.

  1. Our reading

    Language models lack catalog-specific knowledge, so their recommendations tend to be generic and biased toward popular items.

    Their wordslanguage models have world knowledge and can eloquently talk about products, but are unaware of our catalog. Also, their recommendations are generic and suffer from popularity bias.

  2. Our reading

    Traditional recommender systems predict user behavior well but cannot be steered via natural language or explain their own reasoning.

    Their wordsThey excel at predicting what a user will click or buy next, but can't be steered via natural language or reason on their choices.