clip-retrieval
Source: https://github.com/rom1504/clip-retrieval
Structured fields
{
"archived": false,
"revision": {
"ref": "main",
"pinned": false
},
"checked_at": "2026-10-17T00:00:00.000Z",
"usefulness": "MIT toolkit to compute CLIP embeddings for images and text and build a searchable index and web front end; an open route to semantic search over private or open reference-image libraries.",
"source_type": "repository",
"availability": "available",
"dependencies": [
"CLIP / OpenCLIP models"
],
"source_owner": "Romain Beaumont",
"next_check_at": "2026-10-31T00:00:00.000Z",
"hardware_notes": "GPU recommended for embedding at scale (README throughput claim uses a 3080).",
"known_limitations": [
"Throughput claim is upstream's and unmeasured."
],
"upstream_updated_at": "2026-03-28T17:21:35.000Z"
}Evidence claims
- [upstream_claim] README: easily compute CLIP embeddings and build a retrieval system; claims 100M text+image embeddings can be processed in 20 hours on a 3080. · locator: README intro source
- [sourced_fact] Repo metadata: MIT (Copyright 2021 Romain Beaumont), not archived, last push 2026-03-28. · locator: GitHub API fields; LICENSE header source
- [upstream_claim] Research review verdict at export: reviewed (as of 2026-10-17). Recorded by the research pipeline, not a hub review outcome; evidence.review stays unreviewed until an independent reviewer records one.
Rights
| component | license | hosting | redistribution | modification | training | Evidence and notes |
|---|---|---|---|---|---|---|
| code | MIT | permitted | permitted | permitted | unknown | MIT for the tool; embedding model weights it loads carry separate licenses; indexed images keep their own rights. |