{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:WQOEH5XCFFV2MB5G3CZSWH5VJY","short_pith_number":"pith:WQOEH5XC","schema_version":"1.0","canonical_sha256":"b41c43f6e2296ba607a6d8b32b1fb54e267a69b3a60f63e5d326bef7a7f7d0ff","source":{"kind":"arxiv","id":"2104.07204","version":2},"attestation_state":"computed","paper":{"title":"Lattice-BERT: Leveraging Multi-Granularity Representations in Chinese Pre-trained Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Dongyan Zhao, Songfang Huang, Yansong Feng, Yijia Liu, Yuxuan Lai","submitted_at":"2021-04-15T02:36:49Z","abstract_excerpt":"Chinese pre-trained language models usually process text as a sequence of characters, while ignoring more coarse granularity, e.g., words. In this work, we propose a novel pre-training paradigm for Chinese -- Lattice-BERT, which explicitly incorporates word representations along with characters, thus can model a sentence in a multi-granularity manner. Specifically, we construct a lattice graph from the characters and words in a sentence and feed all these text units into transformers. We design a lattice position attention mechanism to exploit the lattice structures in self-attention layers. W"},"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":"2104.07204","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2021-04-15T02:36:49Z","cross_cats_sorted":[],"title_canon_sha256":"724d8912a58f7465421e049196c52ccd655cb0feeb5eade8e12586613286668c","abstract_canon_sha256":"81db396b647758de63e23a26f953932e9c81c6da463267f6abb9c7dd9b43b293"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:44:06.153845Z","signature_b64":"JXR8THu6sfDsvdYScigEVy7Fs5I2F4CIXva48tfkpcUhaeM9kW/m/XaWI05nv9pw++j8+xG+Qe9khSsTHETZAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b41c43f6e2296ba607a6d8b32b1fb54e267a69b3a60f63e5d326bef7a7f7d0ff","last_reissued_at":"2026-07-05T02:44:06.153409Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:44:06.153409Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Lattice-BERT: Leveraging Multi-Granularity Representations in Chinese Pre-trained Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Dongyan Zhao, Songfang Huang, Yansong Feng, Yijia Liu, Yuxuan Lai","submitted_at":"2021-04-15T02:36:49Z","abstract_excerpt":"Chinese pre-trained language models usually process text as a sequence of characters, while ignoring more coarse granularity, e.g., words. In this work, we propose a novel pre-training paradigm for Chinese -- Lattice-BERT, which explicitly incorporates word representations along with characters, thus can model a sentence in a multi-granularity manner. Specifically, we construct a lattice graph from the characters and words in a sentence and feed all these text units into transformers. We design a lattice position attention mechanism to exploit the lattice structures in self-attention layers. W"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2104.07204","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/2104.07204/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":"2104.07204","created_at":"2026-07-05T02:44:06.153465+00:00"},{"alias_kind":"arxiv_version","alias_value":"2104.07204v2","created_at":"2026-07-05T02:44:06.153465+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2104.07204","created_at":"2026-07-05T02:44:06.153465+00:00"},{"alias_kind":"pith_short_12","alias_value":"WQOEH5XCFFV2","created_at":"2026-07-05T02:44:06.153465+00:00"},{"alias_kind":"pith_short_16","alias_value":"WQOEH5XCFFV2MB5G","created_at":"2026-07-05T02:44:06.153465+00:00"},{"alias_kind":"pith_short_8","alias_value":"WQOEH5XC","created_at":"2026-07-05T02:44:06.153465+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2506.07833","citing_title":"Improving Large Language Models with Concept-Aware Fine-Tuning","ref_index":28,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WQOEH5XCFFV2MB5G3CZSWH5VJY","json":"https://pith.science/pith/WQOEH5XCFFV2MB5G3CZSWH5VJY.json","graph_json":"https://pith.science/api/pith-number/WQOEH5XCFFV2MB5G3CZSWH5VJY/graph.json","events_json":"https://pith.science/api/pith-number/WQOEH5XCFFV2MB5G3CZSWH5VJY/events.json","paper":"https://pith.science/paper/WQOEH5XC"},"agent_actions":{"view_html":"https://pith.science/pith/WQOEH5XCFFV2MB5G3CZSWH5VJY","download_json":"https://pith.science/pith/WQOEH5XCFFV2MB5G3CZSWH5VJY.json","view_paper":"https://pith.science/paper/WQOEH5XC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2104.07204&json=true","fetch_graph":"https://pith.science/api/pith-number/WQOEH5XCFFV2MB5G3CZSWH5VJY/graph.json","fetch_events":"https://pith.science/api/pith-number/WQOEH5XCFFV2MB5G3CZSWH5VJY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WQOEH5XCFFV2MB5G3CZSWH5VJY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WQOEH5XCFFV2MB5G3CZSWH5VJY/action/storage_attestation","attest_author":"https://pith.science/pith/WQOEH5XCFFV2MB5G3CZSWH5VJY/action/author_attestation","sign_citation":"https://pith.science/pith/WQOEH5XCFFV2MB5G3CZSWH5VJY/action/citation_signature","submit_replication":"https://pith.science/pith/WQOEH5XCFFV2MB5G3CZSWH5VJY/action/replication_record"}},"created_at":"2026-07-05T02:44:06.153465+00:00","updated_at":"2026-07-05T02:44:06.153465+00:00"}