{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:F63RM22U2UTOC7YEERGG7RP6NV","short_pith_number":"pith:F63RM22U","schema_version":"1.0","canonical_sha256":"2fb7166b54d526e17f04244c6fc5fe6d400a3a6935eaae65b7483a1269edd3c0","source":{"kind":"arxiv","id":"2104.08253","version":2},"attestation_state":"computed","paper":{"title":"Condenser: a Pre-training Architecture for Dense Retrieval","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.IR"],"primary_cat":"cs.CL","authors_text":"Jamie Callan, Luyu Gao","submitted_at":"2021-04-16T17:36:44Z","abstract_excerpt":"Pre-trained Transformer language models (LM) have become go-to text representation encoders. Prior research fine-tunes deep LMs to encode text sequences such as sentences and passages into single dense vector representations for efficient text comparison and retrieval. However, dense encoders require a lot of data and sophisticated techniques to effectively train and suffer in low data situations. This paper finds a key reason is that standard LMs' internal attention structure is not ready-to-use for dense encoders, which needs to aggregate text information into the dense representation. We pr"},"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":false},"canonical_record":{"source":{"id":"2104.08253","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2021-04-16T17:36:44Z","cross_cats_sorted":["cs.IR"],"title_canon_sha256":"7602647d7f4ef92bd13e57889b1d5934e4256ed2d4a895d98c36564a8db0e4f1","abstract_canon_sha256":"b3192b396769516004b57fe85cc48d529b158579284f7edf58bbcc26072c993e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:15:52.472527Z","signature_b64":"ahfuNjtttK5pSw/VIxrj4kXviUf14llxdw8TJztlYMYyKv5rmkTAwV85H+JdmoHXQI9zNPbgL4mH/uG4gIIGDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2fb7166b54d526e17f04244c6fc5fe6d400a3a6935eaae65b7483a1269edd3c0","last_reissued_at":"2026-07-05T03:15:52.472083Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:15:52.472083Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Condenser: a Pre-training Architecture for Dense Retrieval","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.IR"],"primary_cat":"cs.CL","authors_text":"Jamie Callan, Luyu Gao","submitted_at":"2021-04-16T17:36:44Z","abstract_excerpt":"Pre-trained Transformer language models (LM) have become go-to text representation encoders. Prior research fine-tunes deep LMs to encode text sequences such as sentences and passages into single dense vector representations for efficient text comparison and retrieval. However, dense encoders require a lot of data and sophisticated techniques to effectively train and suffer in low data situations. This paper finds a key reason is that standard LMs' internal attention structure is not ready-to-use for dense encoders, which needs to aggregate text information into the dense representation. We pr"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2104.08253","kind":"arxiv","version":2},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2104.08253/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"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":"2104.08253","created_at":"2026-07-05T03:15:52.472128+00:00"},{"alias_kind":"arxiv_version","alias_value":"2104.08253v2","created_at":"2026-07-05T03:15:52.472128+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2104.08253","created_at":"2026-07-05T03:15:52.472128+00:00"},{"alias_kind":"pith_short_12","alias_value":"F63RM22U2UTO","created_at":"2026-07-05T03:15:52.472128+00:00"},{"alias_kind":"pith_short_16","alias_value":"F63RM22U2UTOC7YE","created_at":"2026-07-05T03:15:52.472128+00:00"},{"alias_kind":"pith_short_8","alias_value":"F63RM22U","created_at":"2026-07-05T03:15:52.472128+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2309.07597","citing_title":"C-Pack: Packed Resources For General Chinese Embeddings","ref_index":19,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/F63RM22U2UTOC7YEERGG7RP6NV","json":"https://pith.science/pith/F63RM22U2UTOC7YEERGG7RP6NV.json","graph_json":"https://pith.science/api/pith-number/F63RM22U2UTOC7YEERGG7RP6NV/graph.json","events_json":"https://pith.science/api/pith-number/F63RM22U2UTOC7YEERGG7RP6NV/events.json","paper":"https://pith.science/paper/F63RM22U"},"agent_actions":{"view_html":"https://pith.science/pith/F63RM22U2UTOC7YEERGG7RP6NV","download_json":"https://pith.science/pith/F63RM22U2UTOC7YEERGG7RP6NV.json","view_paper":"https://pith.science/paper/F63RM22U","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2104.08253&json=true","fetch_graph":"https://pith.science/api/pith-number/F63RM22U2UTOC7YEERGG7RP6NV/graph.json","fetch_events":"https://pith.science/api/pith-number/F63RM22U2UTOC7YEERGG7RP6NV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/F63RM22U2UTOC7YEERGG7RP6NV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/F63RM22U2UTOC7YEERGG7RP6NV/action/storage_attestation","attest_author":"https://pith.science/pith/F63RM22U2UTOC7YEERGG7RP6NV/action/author_attestation","sign_citation":"https://pith.science/pith/F63RM22U2UTOC7YEERGG7RP6NV/action/citation_signature","submit_replication":"https://pith.science/pith/F63RM22U2UTOC7YEERGG7RP6NV/action/replication_record"}},"created_at":"2026-07-05T03:15:52.472128+00:00","updated_at":"2026-07-05T03:15:52.472128+00:00"}