Semantic IDs: How to train an LLM-Recommender Hybrid with steerability and reasoning on recommendations. - View it on GitHub