{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:UFNLAAWDDP2EQFCV5JVOYK66EP","short_pith_number":"pith:UFNLAAWD","schema_version":"1.0","canonical_sha256":"a15ab002c31bf4481455ea6aec2bde23e7835b32b7eccc8d1f7c9da9364378a5","source":{"kind":"arxiv","id":"1905.06316","version":1},"attestation_state":"computed","paper":{"title":"What do you learn from context? Probing for sentence structure in contextualized word representations","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Adam Poliak, Alex Wang, Benjamin Van Durme, Berlin Chen, Dipanjan Das, Ellie Pavlick, Ian Tenney, Najoung Kim, Patrick Xia, R Thomas McCoy, Samuel R. Bowman","submitted_at":"2019-05-15T17:48:56Z","abstract_excerpt":"Contextualized representation models such as ELMo (Peters et al., 2018a) and BERT (Devlin et al., 2018) have recently achieved state-of-the-art results on a diverse array of downstream NLP tasks. Building on recent token-level probing work, we introduce a novel edge probing task design and construct a broad suite of sub-sentence tasks derived from the traditional structured NLP pipeline. We probe word-level contextual representations from four recent models and investigate how they encode sentence structure across a range of syntactic, semantic, local, and long-range phenomena. We find that ex"},"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":"1905.06316","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2019-05-15T17:48:56Z","cross_cats_sorted":[],"title_canon_sha256":"e3dc84bf611ef46d0bbfebe0163fa3fac0c053cc5bc73932f4e14e1cf06c006d","abstract_canon_sha256":"5823d090569c65796a1bbd1e30c5f1702c7f31b1d057e24e318be6ba2358a901"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:46:06.864342Z","signature_b64":"BH4fiBRiQreMIUpi34N5Z1bbweG9KuEEwz8cgIwtKzdujW4VbQYA0Oz/HSiBLTW1SZdaZoHq4OeMkDZVetCIDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a15ab002c31bf4481455ea6aec2bde23e7835b32b7eccc8d1f7c9da9364378a5","last_reissued_at":"2026-05-17T23:46:06.863697Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:46:06.863697Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"What do you learn from context? Probing for sentence structure in contextualized word representations","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Adam Poliak, Alex Wang, Benjamin Van Durme, Berlin Chen, Dipanjan Das, Ellie Pavlick, Ian Tenney, Najoung Kim, Patrick Xia, R Thomas McCoy, Samuel R. Bowman","submitted_at":"2019-05-15T17:48:56Z","abstract_excerpt":"Contextualized representation models such as ELMo (Peters et al., 2018a) and BERT (Devlin et al., 2018) have recently achieved state-of-the-art results on a diverse array of downstream NLP tasks. Building on recent token-level probing work, we introduce a novel edge probing task design and construct a broad suite of sub-sentence tasks derived from the traditional structured NLP pipeline. We probe word-level contextual representations from four recent models and investigate how they encode sentence structure across a range of syntactic, semantic, local, and long-range phenomena. We find that ex"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1905.06316","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"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":"1905.06316","created_at":"2026-05-17T23:46:06.863791+00:00"},{"alias_kind":"arxiv_version","alias_value":"1905.06316v1","created_at":"2026-05-17T23:46:06.863791+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1905.06316","created_at":"2026-05-17T23:46:06.863791+00:00"},{"alias_kind":"pith_short_12","alias_value":"UFNLAAWDDP2E","created_at":"2026-05-18T12:33:30.264802+00:00"},{"alias_kind":"pith_short_16","alias_value":"UFNLAAWDDP2EQFCV","created_at":"2026-05-18T12:33:30.264802+00:00"},{"alias_kind":"pith_short_8","alias_value":"UFNLAAWD","created_at":"2026-05-18T12:33:30.264802+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":4,"sample":[{"citing_arxiv_id":"2109.10616","citing_title":"Enriching and Controlling Global Semantics for Text Summarization","ref_index":34,"is_internal_anchor":true},{"citing_arxiv_id":"2605.16378","citing_title":"Mixing Times of Glauber Dynamics on Masked Language Models","ref_index":20,"is_internal_anchor":true},{"citing_arxiv_id":"2602.21750","citing_title":"From Words to Amino Acids: Does the Curse of Depth Persist?","ref_index":35,"is_internal_anchor":true},{"citing_arxiv_id":"2605.12813","citing_title":"REALISTA: Realistic Latent Adversarial Attacks that Elicit LLM Hallucinations","ref_index":22,"is_internal_anchor":true},{"citing_arxiv_id":"2605.05710","citing_title":"On the Blessing of Pre-training in Weak-to-Strong Generalization","ref_index":129,"is_internal_anchor":false},{"citing_arxiv_id":"2604.10786","citing_title":"Do BERT Embeddings Encode Narrative Dimensions? A Token-Level Probing Analysis of Time, Space, Causality, and Character in Fiction","ref_index":3,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/UFNLAAWDDP2EQFCV5JVOYK66EP","json":"https://pith.science/pith/UFNLAAWDDP2EQFCV5JVOYK66EP.json","graph_json":"https://pith.science/api/pith-number/UFNLAAWDDP2EQFCV5JVOYK66EP/graph.json","events_json":"https://pith.science/api/pith-number/UFNLAAWDDP2EQFCV5JVOYK66EP/events.json","paper":"https://pith.science/paper/UFNLAAWD"},"agent_actions":{"view_html":"https://pith.science/pith/UFNLAAWDDP2EQFCV5JVOYK66EP","download_json":"https://pith.science/pith/UFNLAAWDDP2EQFCV5JVOYK66EP.json","view_paper":"https://pith.science/paper/UFNLAAWD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1905.06316&json=true","fetch_graph":"https://pith.science/api/pith-number/UFNLAAWDDP2EQFCV5JVOYK66EP/graph.json","fetch_events":"https://pith.science/api/pith-number/UFNLAAWDDP2EQFCV5JVOYK66EP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UFNLAAWDDP2EQFCV5JVOYK66EP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UFNLAAWDDP2EQFCV5JVOYK66EP/action/storage_attestation","attest_author":"https://pith.science/pith/UFNLAAWDDP2EQFCV5JVOYK66EP/action/author_attestation","sign_citation":"https://pith.science/pith/UFNLAAWDDP2EQFCV5JVOYK66EP/action/citation_signature","submit_replication":"https://pith.science/pith/UFNLAAWDDP2EQFCV5JVOYK66EP/action/replication_record"}},"created_at":"2026-05-17T23:46:06.863791+00:00","updated_at":"2026-05-17T23:46:06.863791+00:00"}