Grevo turns semantic-ID assignment into an evolvable variable for generative recommendation

_reachsumit · x · 2026-07-29

Grevo proposes a unified generative recommendation framework with evolutionary item indexing.
Instead of freezing semantic IDs with a tokenizer, it treats SID assignment as an evolvable discrete variable that adapts to behavioral feedback.
Grevo uses a single multitask recommender to unify behavioral SID generation and semantic SID grounding, absorbing the tokenizer role into the recommender itself.
The system then uses the trained recommender as a posterior evaluator to reassign a budgeted set of high-risk identifiers under a fixed vocabulary budget.

More from coding & agent