{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:OTNLGW4VQPDWVLGUR23MUPBNJN","short_pith_number":"pith:OTNLGW4V","schema_version":"1.0","canonical_sha256":"74dab35b9583c76aacd48eb6ca3c2d4b7f34fddadfa13fdabe3e3960dd2c0be4","source":{"kind":"arxiv","id":"2505.20052","version":1},"attestation_state":"computed","paper":{"title":"Ankh3: Multi-Task Pretraining with Sequence Denoising and Completion Enhances Protein Representations","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["q-bio.QM"],"primary_cat":"cs.LG","authors_text":"Ahmed Elnaggar, Hazem Alsamkary, Mohamed Elkerdawy, Mohamed Elshaffei","submitted_at":"2025-05-26T14:41:10Z","abstract_excerpt":"Protein language models (PLMs) have emerged as powerful tools to detect complex patterns of protein sequences. However, the capability of PLMs to fully capture information on protein sequences might be limited by focusing on single pre-training tasks. Although adding data modalities or supervised objectives can improve the performance of PLMs, pre-training often remains focused on denoising corrupted sequences. To push the boundaries of PLMs, our research investigated a multi-task pre-training strategy. We developed Ankh3, a model jointly optimized on two objectives: masked language modeling 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":"2505.20052","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-26T14:41:10Z","cross_cats_sorted":["q-bio.QM"],"title_canon_sha256":"46d8d7cb334aa26793af3f66ea7e6a021412bbfb693a5e34a5079ed5cdf73ed0","abstract_canon_sha256":"1b0572b8563fdbf00882b1f5d53f5e75f29e8733589ea1f2092a1114265d9c30"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:09:51.909055Z","signature_b64":"cmrKfcRxcfy5Cce2P+qrc6i6gb3U+QWTlDN9C6a00P0dCTRuRsheFcGxr8WajYggYvg+wv/qNWEh8R6Wt+vlAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"74dab35b9583c76aacd48eb6ca3c2d4b7f34fddadfa13fdabe3e3960dd2c0be4","last_reissued_at":"2026-07-05T11:09:51.908565Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:09:51.908565Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Ankh3: Multi-Task Pretraining with Sequence Denoising and Completion Enhances Protein Representations","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["q-bio.QM"],"primary_cat":"cs.LG","authors_text":"Ahmed Elnaggar, Hazem Alsamkary, Mohamed Elkerdawy, Mohamed Elshaffei","submitted_at":"2025-05-26T14:41:10Z","abstract_excerpt":"Protein language models (PLMs) have emerged as powerful tools to detect complex patterns of protein sequences. However, the capability of PLMs to fully capture information on protein sequences might be limited by focusing on single pre-training tasks. Although adding data modalities or supervised objectives can improve the performance of PLMs, pre-training often remains focused on denoising corrupted sequences. To push the boundaries of PLMs, our research investigated a multi-task pre-training strategy. We developed Ankh3, a model jointly optimized on two objectives: masked language modeling w"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.20052","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":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2505.20052/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":"2505.20052","created_at":"2026-07-05T11:09:51.908624+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.20052v1","created_at":"2026-07-05T11:09:51.908624+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.20052","created_at":"2026-07-05T11:09:51.908624+00:00"},{"alias_kind":"pith_short_12","alias_value":"OTNLGW4VQPDW","created_at":"2026-07-05T11:09:51.908624+00:00"},{"alias_kind":"pith_short_16","alias_value":"OTNLGW4VQPDWVLGU","created_at":"2026-07-05T11:09:51.908624+00:00"},{"alias_kind":"pith_short_8","alias_value":"OTNLGW4V","created_at":"2026-07-05T11:09:51.908624+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2603.12845","citing_title":"Multimodal Protein Language Models for Enzyme Kinetic Parameters: From Substrate Recognition to Conformational Adaptation","ref_index":71,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/OTNLGW4VQPDWVLGUR23MUPBNJN","json":"https://pith.science/pith/OTNLGW4VQPDWVLGUR23MUPBNJN.json","graph_json":"https://pith.science/api/pith-number/OTNLGW4VQPDWVLGUR23MUPBNJN/graph.json","events_json":"https://pith.science/api/pith-number/OTNLGW4VQPDWVLGUR23MUPBNJN/events.json","paper":"https://pith.science/paper/OTNLGW4V"},"agent_actions":{"view_html":"https://pith.science/pith/OTNLGW4VQPDWVLGUR23MUPBNJN","download_json":"https://pith.science/pith/OTNLGW4VQPDWVLGUR23MUPBNJN.json","view_paper":"https://pith.science/paper/OTNLGW4V","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.20052&json=true","fetch_graph":"https://pith.science/api/pith-number/OTNLGW4VQPDWVLGUR23MUPBNJN/graph.json","fetch_events":"https://pith.science/api/pith-number/OTNLGW4VQPDWVLGUR23MUPBNJN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OTNLGW4VQPDWVLGUR23MUPBNJN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OTNLGW4VQPDWVLGUR23MUPBNJN/action/storage_attestation","attest_author":"https://pith.science/pith/OTNLGW4VQPDWVLGUR23MUPBNJN/action/author_attestation","sign_citation":"https://pith.science/pith/OTNLGW4VQPDWVLGUR23MUPBNJN/action/citation_signature","submit_replication":"https://pith.science/pith/OTNLGW4VQPDWVLGUR23MUPBNJN/action/replication_record"}},"created_at":"2026-07-05T11:09:51.908624+00:00","updated_at":"2026-07-05T11:09:51.908624+00:00"}