3 ms·
This is achieved using vector databases to store memories as embeddings. Then you can retrieve a “memory” closest to the question in the embedding space.
by coolspot 3y ago
This is achieved using vector databases to store memories as embeddings. Then you can retrieve a “memory” closest to the question in the embedding space.