Skip to content

Ask the HotKV docs

Answers come only from these docs and can be wrong; check the linked pages before you rely on them.

EMBED.DEDUP.PUT

Last updated
EMBED.DEDUP.PUT namespace hash model embedding
Group
Embeddings
Complexity
O(1)
Arity
5 (exactly, counting the command name)
Flags
write, fast
Keys
first 1, last 1, step 1
HotKV Cloud
Performance and Scale
On-Premise
requires the embed-dedup license feature

Embedding de-duplication stores each embedding once per content hash and counts references to it, so the same content embedded by many users or documents is stored, and paid for, only once. You compute the content hash and the embedding; HotKV keeps them in named namespaces. EMBED.DEDUP.PUT stores embedding (any string or binary value, such as packed floats or JSON) under hash with a reference count of 1. If hash is already stored, the stored embedding is kept unchanged and its reference count goes up by 1. model records which embedding model produced it. The namespace is created on first use.

Case RESP2 RESP3
New embedding stored Integer 1 Same
Already stored, reference added Integer 0 Same

Errors:

  • ERR namespace or entry limit reached if a new namespace would exceed max_namespaces, or a new entry would exceed max_entries_per_namespace, under [embed].
  • ERR wrong number of arguments for 'embed.dedup.put' command if the argument count is wrong.
  • NOLICENSE embed-dedup requires a HotKV license. Contact sales@hotkv.com if your license does not include the embed-dedup feature.
127.0.0.1:6379> EMBED.DEDUP.PUT org abc123 text-embedding-3-small "vector-bytes"
(integer) 1
127.0.0.1:6379> EMBED.DEDUP.PUT org abc123 text-embedding-3-small "vector-bytes"
(integer) 0

HotKV-only command. Redis and Valkey have no equivalent.

EMBED.DEDUP.GET, EMBED.DEDUP.RELEASE, EMBED.DEDUP.STATS, EMBED.DEDUP.PURGE, EMBED.DEDUP.DROP, EMBED.DEDUP.LIST, AI features guide