{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:Z52BXLJPOVYNXCWOMA3GFJLDTO","short_pith_number":"pith:Z52BXLJP","schema_version":"1.0","canonical_sha256":"cf741bad2f7570db8ace603662a5639bbdb476bcbc363e216b26c2d777f1a6ac","source":{"kind":"arxiv","id":"2210.07316","version":3},"attestation_state":"computed","paper":{"title":"MTEB: Massive Text Embedding Benchmark","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A new benchmark shows no single text embedding method performs best across all tasks.","cross_cats":["cs.IR","cs.LG"],"primary_cat":"cs.CL","authors_text":"Lo\\\"ic Magne, Niklas Muennighoff, Nils Reimers, Nouamane Tazi","submitted_at":"2022-10-13T19:42:08Z","abstract_excerpt":"Text embeddings are commonly evaluated on a small set of datasets from a single task not covering their possible applications to other tasks. It is unclear whether state-of-the-art embeddings on semantic textual similarity (STS) can be equally well applied to other tasks like clustering or reranking. This makes progress in the field difficult to track, as various models are constantly being proposed without proper evaluation. To solve this problem, we introduce the Massive Text Embedding Benchmark (MTEB). MTEB spans 8 embedding tasks covering a total of 58 datasets and 112 languages. Through t"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":true},"canonical_record":{"source":{"id":"2210.07316","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2022-10-13T19:42:08Z","cross_cats_sorted":["cs.IR","cs.LG"],"title_canon_sha256":"c6029532035edb5cd95cc52d76c8c664a42e23bb7c566963351eb1aac851f070","abstract_canon_sha256":"b9f0eef31e7d1249bfde4e27f63923dc5697cd582fa9ac5ce1b60fe9dea003f8"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:38:52.800465Z","signature_b64":"k8355wyFAxIYh25I/ji7qG/Re7KE2M1dTL03wjYGfsj4Crs9MmQpUiFdF9a9vl/SerK6/1TrLxNjD1pTO6pyAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cf741bad2f7570db8ace603662a5639bbdb476bcbc363e216b26c2d777f1a6ac","last_reissued_at":"2026-05-17T23:38:52.799771Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:38:52.799771Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MTEB: Massive Text Embedding Benchmark","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A new benchmark shows no single text embedding method performs best across all tasks.","cross_cats":["cs.IR","cs.LG"],"primary_cat":"cs.CL","authors_text":"Lo\\\"ic Magne, Niklas Muennighoff, Nils Reimers, Nouamane Tazi","submitted_at":"2022-10-13T19:42:08Z","abstract_excerpt":"Text embeddings are commonly evaluated on a small set of datasets from a single task not covering their possible applications to other tasks. It is unclear whether state-of-the-art embeddings on semantic textual similarity (STS) can be equally well applied to other tasks like clustering or reranking. This makes progress in the field difficult to track, as various models are constantly being proposed without proper evaluation. To solve this problem, we introduce the Massive Text Embedding Benchmark (MTEB). MTEB spans 8 embedding tasks covering a total of 58 datasets and 112 languages. Through t"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Through the benchmarking of 33 models on MTEB, we establish the most comprehensive benchmark of text embeddings to date. We find that no particular text embedding method dominates across all tasks.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The 8 tasks and 58 datasets chosen are sufficiently representative of the full range of real-world embedding applications so that performance on MTEB predicts usefulness elsewhere.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"MTEB is a new multi-task benchmark showing that no single text embedding model leads on all 8 tasks across 58 datasets.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A new benchmark shows no single text embedding method performs best across all tasks.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"7ab1a36f952b2168275f60602fb2ab1252318161184fa61c3159a31bda5c3ac4"},"source":{"id":"2210.07316","kind":"arxiv","version":3},"verdict":{"id":"26569f9a-67bb-44a6-a5cd-e1af2c709a23","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T10:11:33.391781Z","strongest_claim":"Through the benchmarking of 33 models on MTEB, we establish the most comprehensive benchmark of text embeddings to date. We find that no particular text embedding method dominates across all tasks.","one_line_summary":"MTEB is a new multi-task benchmark showing that no single text embedding model leads on all 8 tasks across 58 datasets.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The 8 tasks and 58 datasets chosen are sufficiently representative of the full range of real-world embedding applications so that performance on MTEB predicts usefulness elsewhere.","pith_extraction_headline":"A new benchmark shows no single text embedding method performs best across all tasks."},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":1,"snapshot_sha256":"dd497eacc74514a675723cf8d1755fc838ee4ed8793d9c1c4f117c64022fad16"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2210.07316","created_at":"2026-05-17T23:38:52.799871+00:00"},{"alias_kind":"arxiv_version","alias_value":"2210.07316v3","created_at":"2026-05-17T23:38:52.799871+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2210.07316","created_at":"2026-05-17T23:38:52.799871+00:00"},{"alias_kind":"pith_short_12","alias_value":"Z52BXLJPOVYN","created_at":"2026-05-18T12:33:33.725879+00:00"},{"alias_kind":"pith_short_16","alias_value":"Z52BXLJPOVYNXCWO","created_at":"2026-05-18T12:33:33.725879+00:00"},{"alias_kind":"pith_short_8","alias_value":"Z52BXLJP","created_at":"2026-05-18T12:33:33.725879+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":40,"internal_anchor_count":40,"sample":[{"citing_arxiv_id":"2412.14751","citing_title":"Query pipeline optimization for cancer patient question answering systems","ref_index":45,"is_internal_anchor":true},{"citing_arxiv_id":"2605.22544","citing_title":"One prompt is not enough: Instruction Sensitivity Undermines Embedding Model Evaluation","ref_index":23,"is_internal_anchor":true},{"citing_arxiv_id":"2507.03122","citing_title":"Federated Learning for ICD Classification with Lightweight Models and Pretrained Embeddings","ref_index":19,"is_internal_anchor":true},{"citing_arxiv_id":"2602.06358","citing_title":"SHINE: A Scalable In-Context Hypernetwork for Mapping Context to LoRA in a Single Pass","ref_index":15,"is_internal_anchor":true},{"citing_arxiv_id":"2605.20268","citing_title":"Chronicle: A Multimodal Foundation Model for Joint Language and Time Series Understanding","ref_index":21,"is_internal_anchor":true},{"citing_arxiv_id":"2605.20724","citing_title":"CALMem : Application-Layer Dual Memory for Conversational AI","ref_index":6,"is_internal_anchor":true},{"citing_arxiv_id":"2509.08150","citing_title":"Verbalized Algorithms: Classical Algorithms are All You Need (Mostly)","ref_index":10,"is_internal_anchor":true},{"citing_arxiv_id":"2510.01706","citing_title":"Representational Alignment Across Model Layers and Brain Regions with Multi-Level Optimal Transport","ref_index":10,"is_internal_anchor":true},{"citing_arxiv_id":"2510.01801","citing_title":"Detecting LLM-Generated Spam Reviews by Integrating Language Model Embeddings and Graph Neural Network","ref_index":30,"is_internal_anchor":true},{"citing_arxiv_id":"2510.07048","citing_title":"Search-R3: Unifying Reasoning and Embedding in Large Language Models","ref_index":49,"is_internal_anchor":true},{"citing_arxiv_id":"2305.16264","citing_title":"Scaling Data-Constrained Language Models","ref_index":79,"is_internal_anchor":true},{"citing_arxiv_id":"2511.07969","citing_title":"Unified Work Embeddings: Contrastive Learning of a Bidirectional Multi-task Ranker","ref_index":30,"is_internal_anchor":true},{"citing_arxiv_id":"2511.11010","citing_title":"GovScape: A Public Multimodal Search System for 70 Million Pages of Government PDFs","ref_index":37,"is_internal_anchor":true},{"citing_arxiv_id":"2511.15424","citing_title":"LLM-MemCluster: Empowering Large Language Models with Dynamic Memory for Text Clustering","ref_index":10,"is_internal_anchor":true},{"citing_arxiv_id":"2301.12652","citing_title":"REPLUG: Retrieval-Augmented Black-Box Language Models","ref_index":64,"is_internal_anchor":true},{"citing_arxiv_id":"2512.14067","citing_title":"Efficient-DLM: From Autoregressive to Diffusion Language Models, and Beyond in Speed","ref_index":23,"is_internal_anchor":true},{"citing_arxiv_id":"2512.16182","citing_title":"DualGuard: Dual-stream Large Language Model Watermarking Defense against Paraphrase and Spoofing Attack","ref_index":41,"is_internal_anchor":true},{"citing_arxiv_id":"2604.06176","citing_title":"Robustness Risk of Conversational Retrieval: Identifying and Mitigating Noise Sensitivity in Qwen3-Embedding Model","ref_index":5,"is_internal_anchor":true},{"citing_arxiv_id":"2602.22611","citing_title":"Mitigating Membership Inference in Intermediate Representations with Differentially Private Training","ref_index":8,"is_internal_anchor":true},{"citing_arxiv_id":"2509.20354","citing_title":"EmbeddingGemma: Powerful and Lightweight Text Representations","ref_index":15,"is_internal_anchor":true},{"citing_arxiv_id":"2604.13056","citing_title":"Text-as-Signal: Quantitative Semantic Scoring with Embeddings, Logprobs, and Noise Reduction","ref_index":4,"is_internal_anchor":true},{"citing_arxiv_id":"2405.17428","citing_title":"NV-Embed: Improved Techniques for Training LLMs as Generalist Embedding Models","ref_index":115,"is_internal_anchor":true},{"citing_arxiv_id":"2605.13149","citing_title":"AcquisitionSynthesis: Targeted Data Generation using Acquisition Functions","ref_index":28,"is_internal_anchor":true},{"citing_arxiv_id":"2604.02655","citing_title":"Semantic Data Processing with Holistic Data Understanding","ref_index":45,"is_internal_anchor":true},{"citing_arxiv_id":"2309.07597","citing_title":"C-Pack: Packed Resources For General Chinese Embeddings","ref_index":38,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":1,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/Z52BXLJPOVYNXCWOMA3GFJLDTO","json":"https://pith.science/pith/Z52BXLJPOVYNXCWOMA3GFJLDTO.json","graph_json":"https://pith.science/api/pith-number/Z52BXLJPOVYNXCWOMA3GFJLDTO/graph.json","events_json":"https://pith.science/api/pith-number/Z52BXLJPOVYNXCWOMA3GFJLDTO/events.json","paper":"https://pith.science/paper/Z52BXLJP"},"agent_actions":{"view_html":"https://pith.science/pith/Z52BXLJPOVYNXCWOMA3GFJLDTO","download_json":"https://pith.science/pith/Z52BXLJPOVYNXCWOMA3GFJLDTO.json","view_paper":"https://pith.science/paper/Z52BXLJP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2210.07316&json=true","fetch_graph":"https://pith.science/api/pith-number/Z52BXLJPOVYNXCWOMA3GFJLDTO/graph.json","fetch_events":"https://pith.science/api/pith-number/Z52BXLJPOVYNXCWOMA3GFJLDTO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/Z52BXLJPOVYNXCWOMA3GFJLDTO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/Z52BXLJPOVYNXCWOMA3GFJLDTO/action/storage_attestation","attest_author":"https://pith.science/pith/Z52BXLJPOVYNXCWOMA3GFJLDTO/action/author_attestation","sign_citation":"https://pith.science/pith/Z52BXLJPOVYNXCWOMA3GFJLDTO/action/citation_signature","submit_replication":"https://pith.science/pith/Z52BXLJPOVYNXCWOMA3GFJLDTO/action/replication_record"}},"created_at":"2026-05-17T23:38:52.799871+00:00","updated_at":"2026-05-17T23:38:52.799871+00:00"}