Equivalency-graphs could be promising, see [babble](https://arxiv.org/abs/2212.04596). Merkle-tree deduplication will still be useful even with library learning.
Equivalency-graphs could be promising, see babble.
Merkle-tree deduplication will still be useful even with library learning.