{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2020:3EPBZXRXEYBRKDI24UMSXX2YWE","short_pith_number":"pith:3EPBZXRX","canonical_record":{"source":{"id":"2006.16362","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-06-29T20:28:52Z","cross_cats_sorted":["cs.CL","stat.ML"],"title_canon_sha256":"0508f76419d658e9105ee55c2dd9444d6d9a8dd0426f2f1d3481c5ffdacb0b5e","abstract_canon_sha256":"507b8b7c67bda8c2bd89f9076601a71697af4e27fbb775f681f73e65d9e1bd94"},"schema_version":"1.0"},"canonical_sha256":"d91e1cde372603150d1ae5192bdf58b1320d95e28727fa4fcf41a7fde5bfaabd","source":{"kind":"arxiv","id":"2006.16362","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2006.16362","created_at":"2026-07-05T02:41:47Z"},{"alias_kind":"arxiv_version","alias_value":"2006.16362v2","created_at":"2026-07-05T02:41:47Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2006.16362","created_at":"2026-07-05T02:41:47Z"},{"alias_kind":"pith_short_12","alias_value":"3EPBZXRXEYBR","created_at":"2026-07-05T02:41:47Z"},{"alias_kind":"pith_short_16","alias_value":"3EPBZXRXEYBRKDI2","created_at":"2026-07-05T02:41:47Z"},{"alias_kind":"pith_short_8","alias_value":"3EPBZXRX","created_at":"2026-07-05T02:41:47Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2020:3EPBZXRXEYBRKDI24UMSXX2YWE","target":"record","payload":{"canonical_record":{"source":{"id":"2006.16362","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-06-29T20:28:52Z","cross_cats_sorted":["cs.CL","stat.ML"],"title_canon_sha256":"0508f76419d658e9105ee55c2dd9444d6d9a8dd0426f2f1d3481c5ffdacb0b5e","abstract_canon_sha256":"507b8b7c67bda8c2bd89f9076601a71697af4e27fbb775f681f73e65d9e1bd94"},"schema_version":"1.0"},"canonical_sha256":"d91e1cde372603150d1ae5192bdf58b1320d95e28727fa4fcf41a7fde5bfaabd","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:41:47.349887Z","signature_b64":"CqxYsf+0JNhTpCZx9CTs+DRcqiVC69+X6BJPQRKp/QXO2LMhsYK+Zr6b8UgYpUjwKGlPUhotBeHKo5h3AmEBDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d91e1cde372603150d1ae5192bdf58b1320d95e28727fa4fcf41a7fde5bfaabd","last_reissued_at":"2026-07-05T02:41:47.349339Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:41:47.349339Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2006.16362","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T02:41:47Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"ZBBNw98n1uH5jhl0P3Yx1X5CzaRtJRwLCRR1MXo/fi2fXoUOqe6aGJwyj1KS1rnj8YT+3JftLwFSeVc8B0aFCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-05T02:09:58.463899Z"},"content_sha256":"93c7739aee44a82d3c4920437297675be304d62af07e207986fca3ea904170bd","schema_version":"1.0","event_id":"sha256:93c7739aee44a82d3c4920437297675be304d62af07e207986fca3ea904170bd"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2020:3EPBZXRXEYBRKDI24UMSXX2YWE","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Multi-Head Attention: Collaborate Instead of Concatenate","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Andreas Loukas, Jean-Baptiste Cordonnier, Martin Jaggi","submitted_at":"2020-06-29T20:28:52Z","abstract_excerpt":"Attention layers are widely used in natural language processing (NLP) and are beginning to influence computer vision architectures. Training very large transformer models allowed significant improvement in both fields, but once trained, these networks show symptoms of over-parameterization. For instance, it is known that many attention heads can be pruned without impacting accuracy. This work aims to enhance current understanding on how multiple heads interact. Motivated by the observation that attention heads learn redundant key/query projections, we propose a collaborative multi-head attenti"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2006.16362","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/2006.16362/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T02:41:47Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"v0ZzJjwzfCWBfqx87RC4FD3RDDTwzdFAAAQtJxZG2Z5fyCkN79ruMrnVJVVOY6SuiG0OMuTulazf0GkT7XctAg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-05T02:09:58.464273Z"},"content_sha256":"523e9961fe8e8d92196e2f55c64551efd74522b2c9a232a10f24fdc7f8535956","schema_version":"1.0","event_id":"sha256:523e9961fe8e8d92196e2f55c64551efd74522b2c9a232a10f24fdc7f8535956"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/3EPBZXRXEYBRKDI24UMSXX2YWE/bundle.json","state_url":"https://pith.science/pith/3EPBZXRXEYBRKDI24UMSXX2YWE/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/3EPBZXRXEYBRKDI24UMSXX2YWE/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-05T02:09:58Z","links":{"resolver":"https://pith.science/pith/3EPBZXRXEYBRKDI24UMSXX2YWE","bundle":"https://pith.science/pith/3EPBZXRXEYBRKDI24UMSXX2YWE/bundle.json","state":"https://pith.science/pith/3EPBZXRXEYBRKDI24UMSXX2YWE/state.json","well_known_bundle":"https://pith.science/.well-known/pith/3EPBZXRXEYBRKDI24UMSXX2YWE/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2020:3EPBZXRXEYBRKDI24UMSXX2YWE","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":"507b8b7c67bda8c2bd89f9076601a71697af4e27fbb775f681f73e65d9e1bd94","cross_cats_sorted":["cs.CL","stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-06-29T20:28:52Z","title_canon_sha256":"0508f76419d658e9105ee55c2dd9444d6d9a8dd0426f2f1d3481c5ffdacb0b5e"},"schema_version":"1.0","source":{"id":"2006.16362","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2006.16362","created_at":"2026-07-05T02:41:47Z"},{"alias_kind":"arxiv_version","alias_value":"2006.16362v2","created_at":"2026-07-05T02:41:47Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2006.16362","created_at":"2026-07-05T02:41:47Z"},{"alias_kind":"pith_short_12","alias_value":"3EPBZXRXEYBR","created_at":"2026-07-05T02:41:47Z"},{"alias_kind":"pith_short_16","alias_value":"3EPBZXRXEYBRKDI2","created_at":"2026-07-05T02:41:47Z"},{"alias_kind":"pith_short_8","alias_value":"3EPBZXRX","created_at":"2026-07-05T02:41:47Z"}],"graph_snapshots":[{"event_id":"sha256:523e9961fe8e8d92196e2f55c64551efd74522b2c9a232a10f24fdc7f8535956","target":"graph","created_at":"2026-07-05T02:41:47Z","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/2006.16362/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Attention layers are widely used in natural language processing (NLP) and are beginning to influence computer vision architectures. Training very large transformer models allowed significant improvement in both fields, but once trained, these networks show symptoms of over-parameterization. For instance, it is known that many attention heads can be pruned without impacting accuracy. This work aims to enhance current understanding on how multiple heads interact. Motivated by the observation that attention heads learn redundant key/query projections, we propose a collaborative multi-head attenti","authors_text":"Andreas Loukas, Jean-Baptiste Cordonnier, Martin Jaggi","cross_cats":["cs.CL","stat.ML"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-06-29T20:28:52Z","title":"Multi-Head Attention: Collaborate Instead of Concatenate"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2006.16362","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":"sha256:93c7739aee44a82d3c4920437297675be304d62af07e207986fca3ea904170bd","target":"record","created_at":"2026-07-05T02:41:47Z","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":"507b8b7c67bda8c2bd89f9076601a71697af4e27fbb775f681f73e65d9e1bd94","cross_cats_sorted":["cs.CL","stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-06-29T20:28:52Z","title_canon_sha256":"0508f76419d658e9105ee55c2dd9444d6d9a8dd0426f2f1d3481c5ffdacb0b5e"},"schema_version":"1.0","source":{"id":"2006.16362","kind":"arxiv","version":2}},"canonical_sha256":"d91e1cde372603150d1ae5192bdf58b1320d95e28727fa4fcf41a7fde5bfaabd","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"d91e1cde372603150d1ae5192bdf58b1320d95e28727fa4fcf41a7fde5bfaabd","first_computed_at":"2026-07-05T02:41:47.349339Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T02:41:47.349339Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"CqxYsf+0JNhTpCZx9CTs+DRcqiVC69+X6BJPQRKp/QXO2LMhsYK+Zr6b8UgYpUjwKGlPUhotBeHKo5h3AmEBDg==","signature_status":"signed_v1","signed_at":"2026-07-05T02:41:47.349887Z","signed_message":"canonical_sha256_bytes"},"source_id":"2006.16362","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:93c7739aee44a82d3c4920437297675be304d62af07e207986fca3ea904170bd","sha256:523e9961fe8e8d92196e2f55c64551efd74522b2c9a232a10f24fdc7f8535956"],"state_sha256":"4bbae36cde70aae8e4c826b8e92642e47fd53c244975dc8d7ddaa24b40c233b0"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Thx/LVVghrpIIGfmcUERIKZZiEip3HdbCmdop8sergdYCJVBwO9aic18V1p7Tkjv9Ql6diYCxIKwdozUpporDQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-05T02:09:58.467471Z","bundle_sha256":"cedb6c234a6cb7de82067067fe220f20ad502a137851312e85c97bb52c409dfb"}}