Atlas
https://arxiv.org/abs/2208.03299
Meta AI. Jointly fine-tunes retriever + LM for few-shot tasks. The facebookresearch/atlas repo is no longer maintained; the code is provided as is.
Research code · Verified 2026-10-29 · fresh as of 2026-10-29 · 32 cited facts, 1 with a third-party source · 2 recorded changes · Suggest a correction
At a glance
- Type
- Few-shot retrieval-augmented LM vendor fix
- Tier
- T3
- Section
- Retrieval-as-memory hybrids
- License
- CC-BY-NC 4.0 repo fix
- GitHub
- facebookresearch/atlas — ~560 stars, unmaintained repo fix
- Pricing
- searched not found fix
- Funding
- not applicable — not commercial fix
- Freshness
- 2026-10-29 fix
- Layer
- agent-infra repo fix
- Licensing
- commercial repo fix
- Developer activity
- OpenRank 0.58 (12-mo avg 1.5) · 1 participants — 2024-08 third-party fix
- Maintainer health
- 0 issues in 90 days · 0 releases in 12 months · bus factor 1 — as of 2026-10-31 repo
Memory taxonomy
- storage
- vector
- retrieval
- similarity
- persistence
- long-term
- update
- parametric-edit
- unit
- document
- governance
- inspectable
- conflict
- n/a
When to use
Optimised for: retrieval quality / accuracy on long-context QA
Anti-fit: most are research papers; production fitness varies by maintainer
Pros & cons
Pros
Meta retrieval architecture for few-shot learning.
Cons
Research scope; less production utility than fine-tuned RAG.
Claims & capabilities
Long-context architecture optimizing recall via test-time memory. Headline: +80% accuracy at 10M-token BABILong vs. Transformers and recent linear-recurrent baselines; baselines: Transformers and linear-recurrent models; primary dataset: BABILong (plus language modeling, common-sense reasoning, recall-intensive long-context tasks).
Technical surface
- API surface
- not applicable — research paper
- Backend storage
- not applicable — research paper
- Deployment
- searched not found
- Embedding model
- not applicable — research paper
- Multi-tenancy
- not applicable — research paper
- MCP
- not applicable — research paper, no deployed product
- A2A
- not applicable — research paper, no deployed product
- OpenTelemetry
- not applicable — research paper, no deployed product
Alternatives to Atlas → live options ranked by shared design, licence, hosting and MCP support
Similar systems
Other retrieval-as-memory hybrids in the catalog, ranked by inbound references.
- GraphRAG (Microsoft) T1
Modular graph-based RAG. Leiden community detection over knowledge graph; community summaries at multiple levels. DRIFT Search hybrid (Oct 2024). The GraphRAG repo is now largely in maintenance mode: no new features or PRs, only bug fixes and dependency/CVE updates.
- RETRO T3
DeepMind. Retrieval-Enhanced Transformer with chunked cross-attention over a 2T-token retrieval database. Frozen BERT retriever + differentiable encoder.
- RAPTOR T3
Stanford recursive-summarisation tree over the corpus — clusters and summarises documents into a hierarchical index; retrieves at multiple levels. ICLR 2024.
- Self-RAG T4
Trains an LLM end-to-end to emit reflection tokens that decide whether to retrieve, assess passage relevance, and critique output quality — all within a single model.
- HippoRAG / HippoRAG2 T3
Hippocampus-inspired RAG. LLMs + knowledge graphs + Personalized PageRank to mimic neocortex/hippocampus roles. NeurIPS 2024.
- MemoRAG T4
RAG variant with a long-context global memory model that drafts answer clues used to guide passage retrieval — improves recall in noisy multi-hop QA.
Sources (31 cited facts, 1 third-party)
Each fact links to the page it came from; the badge says who published it (how to read sources).
- What it is Few-shot retrieval-augmented LM vendor
- What it does & what's distinct Meta AI. Jointly fine-tunes retriever + LM for few-shot tasks. The facebookresearch/atla… repo
- Claims / benefits Long-context architecture optimizing recall via test-time memory. Headline: +80% accurac… vendor
- Modalities text vendor
- Created 2022-11 vendor
- Latest release n/a repo
- License CC-BY-NC 4.0 repo
- GitHub facebookresearch/atlas — ~560 stars, unmaintained repo
- Mindshare 42 cites · 3.7/mo paper
- Citations 1.2k cites 296/yr paper
- Founders / pedigree Meta AI; arxiv 2208.03299 vendor
- Memory volume / scale 26.8M Wikipedia documents (research corpus) vendor
- Performance +80% acc BABILong vendor
- Reproducibility Independently reproduced: engineerA314/atlas-pytorch (unofficial PyTorch port; no offici… repo
- Code/weights release Code public + weights public (pretrained base/large/xl/xxl checkpoints); official repo f… repo
- Versioning snapshots vendor
- Schema evolution schema with migration vendor
- Validated verticals research, RAG developers (horizontal); some have production deployments vendor
- Anti-fit most are research papers; production fitness varies by maintainer vendor
- Optimised for retrieval quality / accuracy on long-context QA vendor
- Adjacent infrastructure BYO LLM + vector store; OSS reference implementations exist vendor
- Pros Meta retrieval architecture for few-shot learning. vendor
- Cons Research scope; less production utility than fine-tuned RAG. vendor
- Project & sources arxiv 2208.03299 vendor
- commit-trajectory [{"ym":"2022-11","cum":3},{"ym":"2022-12","cum":8},{"ym":"2023-04","cum":10},{"ym":"2023… repo
- citation-trajectory [{"year":2023,"cum":6,"infl":2},{"year":2024,"cum":14,"infl":3},{"year":2025,"cum":18,"i… paper
- architecture-deployment-locus local repo
- licensing-bucket commercial repo
- positioning-platform-vs-product platform repo
- taxonomy-three-layer-rollup agent-infra repo
- developer-activity-opendigger {"activity":1.49,"as_of":"2024-08","openrank_12mo":1.5,"openrank_current":0.58,"particip… third-party
What changed
Changes to this record's lifecycle, licence, release, pricing and hosting facts, from the catalog's git history. Subscribe (Atom)
Related systems
References (1)
- RETRO cites — S2 isInfluential citation
Referenced by (5)
- ITER-RETGEN cites — S2 isInfluential citation
- LongRAG cites — S2 isInfluential citation
- MemBART cites — S2 isInfluential citation
- Memformers (gradient memory) cites — S2 isInfluential citation
- Toolformer cites — S2 isInfluential citation