{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:RJAY4LK65VA2UI5OKGOE4AA6PT","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"4836bb73b9ec5200599aa4d76147eb01b9b387c8bfcb6591fe2f4cea11a237de","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.LG","submitted_at":"2024-06-03T13:28:43Z","title_canon_sha256":"c8051f3ad2ded372c2ac7c54337756d871b90e6b1405e63bc22d5277511732df"},"schema_version":"1.0","source":{"id":"2406.06567","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2406.06567","created_at":"2026-07-05T09:45:50Z"},{"alias_kind":"arxiv_version","alias_value":"2406.06567v2","created_at":"2026-07-05T09:45:50Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.06567","created_at":"2026-07-05T09:45:50Z"},{"alias_kind":"pith_short_12","alias_value":"RJAY4LK65VA2","created_at":"2026-07-05T09:45:50Z"},{"alias_kind":"pith_short_16","alias_value":"RJAY4LK65VA2UI5O","created_at":"2026-07-05T09:45:50Z"},{"alias_kind":"pith_short_8","alias_value":"RJAY4LK6","created_at":"2026-07-05T09:45:50Z"}],"graph_snapshots":[{"event_id":"sha256:b255a1d93d4b9eb8ec86d257c4a4d531ead8470337a16eb19f099ba77109a4a5","target":"graph","created_at":"2026-07-05T09:45:50Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2406.06567/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large language models (LLMs) with billions of parameters demonstrate impressive performance. However, the widely used Multi-Head Attention (MHA) in LLMs incurs substantial computational and memory costs during inference. While some efforts have optimized attention mechanisms by pruning heads or sharing parameters among heads, these methods often lead to performance degradation or necessitate substantial continued pre-training costs to restore performance. Based on the analysis of attention redundancy, we design a Decoupled-Head Attention (DHA) mechanism. DHA adaptively configures group sharing","authors_text":"Junyuan Shang, Linhao Zhang, Shuohuan Wang, Tingwen Liu, Yilong Chen, Yu Sun, Zhenyu Zhang","cross_cats":["cs.AI","cs.CL"],"headline":"","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.LG","submitted_at":"2024-06-03T13:28:43Z","title":"DHA: Learning Decoupled-Head Attention from Transformer Checkpoints via Adaptive Heads Fusion"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.06567","kind":"arxiv","version":2},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:4f73e9bc64dba5dec2e94630d455d8acbacfb984b3d86f2be5f6eeeed922c304","target":"record","created_at":"2026-07-05T09:45:50Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"4836bb73b9ec5200599aa4d76147eb01b9b387c8bfcb6591fe2f4cea11a237de","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.LG","submitted_at":"2024-06-03T13:28:43Z","title_canon_sha256":"c8051f3ad2ded372c2ac7c54337756d871b90e6b1405e63bc22d5277511732df"},"schema_version":"1.0","source":{"id":"2406.06567","kind":"arxiv","version":2}},"canonical_sha256":"8a418e2d5eed41aa23ae519c4e001e7cf80e9d32ab0905c42d604138702322d6","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"8a418e2d5eed41aa23ae519c4e001e7cf80e9d32ab0905c42d604138702322d6","first_computed_at":"2026-07-05T09:45:50.079796Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T09:45:50.079796Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"m+FH66AI0+zj0b8S5UtZcd7JHzbOUTKtO0p7Q3Gtmm6ZWZqbXkCvcpuZa7oN8yF58M1wgfj9YWMm01S2fUDVDw==","signature_status":"signed_v1","signed_at":"2026-07-05T09:45:50.080271Z","signed_message":"canonical_sha256_bytes"},"source_id":"2406.06567","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:4f73e9bc64dba5dec2e94630d455d8acbacfb984b3d86f2be5f6eeeed922c304","sha256:b255a1d93d4b9eb8ec86d257c4a4d531ead8470337a16eb19f099ba77109a4a5"],"state_sha256":"c652f85bbc72a9a0af3c01c7efd4aef9ee7edcfbd0ef991c3350ae95bfeb8c39"}