{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2021:EZHUFD3ZJX7ZAOEN4BMKSVNHR2","short_pith_number":"pith:EZHUFD3Z","canonical_record":{"source":{"id":"2110.14048","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-10-26T22:03:51Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"143861749e0631ad1931e3db718686390ab8a5b340606a75b7c30f9eb4a4b74a","abstract_canon_sha256":"63ae58c9d996e91e42726851562e781429d169e8d37e9be58934729cc0d059ec"},"schema_version":"1.0"},"canonical_sha256":"264f428f794dff90388de058a955a78e85f4348332cb1ceaa5f125aadb34d648","source":{"kind":"arxiv","id":"2110.14048","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2110.14048","created_at":"2026-07-05T07:47:23Z"},{"alias_kind":"arxiv_version","alias_value":"2110.14048v2","created_at":"2026-07-05T07:47:23Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2110.14048","created_at":"2026-07-05T07:47:23Z"},{"alias_kind":"pith_short_12","alias_value":"EZHUFD3ZJX7Z","created_at":"2026-07-05T07:47:23Z"},{"alias_kind":"pith_short_16","alias_value":"EZHUFD3ZJX7ZAOEN","created_at":"2026-07-05T07:47:23Z"},{"alias_kind":"pith_short_8","alias_value":"EZHUFD3Z","created_at":"2026-07-05T07:47:23Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2021:EZHUFD3ZJX7ZAOEN4BMKSVNHR2","target":"record","payload":{"canonical_record":{"source":{"id":"2110.14048","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-10-26T22:03:51Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"143861749e0631ad1931e3db718686390ab8a5b340606a75b7c30f9eb4a4b74a","abstract_canon_sha256":"63ae58c9d996e91e42726851562e781429d169e8d37e9be58934729cc0d059ec"},"schema_version":"1.0"},"canonical_sha256":"264f428f794dff90388de058a955a78e85f4348332cb1ceaa5f125aadb34d648","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:47:23.361645Z","signature_b64":"BXcKl1Vuhm9JV6iirCGMbqSfET2Vqc+CXN01jmcFC3JnoGPibr0ME42+/WjSiPgONpMg9YhFwTBvajBsOVZ1Bg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"264f428f794dff90388de058a955a78e85f4348332cb1ceaa5f125aadb34d648","last_reissued_at":"2026-07-05T07:47:23.361154Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:47:23.361154Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2110.14048","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":"2026-07-05T07:47:23Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"YV7UQAxdCIN3NogfgSE8mlB7jVLMRQ242xZYEW1bQgrOBIQ9hQV1O2xZ6tBVARrxxrdxsVIG1TMD5ZT8ytx5Bg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-17T01:47:52.600761Z"},"content_sha256":"a5ecdaf494954019b180379a3bb5f2be531195f98d8a8fd00455a68e95c959d9","schema_version":"1.0","event_id":"sha256:a5ecdaf494954019b180379a3bb5f2be531195f98d8a8fd00455a68e95c959d9"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2021:EZHUFD3ZJX7ZAOEN4BMKSVNHR2","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Conflict-Averse Gradient Descent for Multi-task Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Bo Liu, Peter Stone, Qiang Liu, Xiaojie Jin, Xingchao Liu","submitted_at":"2021-10-26T22:03:51Z","abstract_excerpt":"The goal of multi-task learning is to enable more efficient learning than single task learning by sharing model structures for a diverse set of tasks. A standard multi-task learning objective is to minimize the average loss across all tasks. While straightforward, using this objective often results in much worse final performance for each task than learning them independently. A major challenge in optimizing a multi-task model is the conflicting gradients, where gradients of different task objectives are not well aligned so that following the average gradient direction can be detrimental to sp"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2110.14048","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/2110.14048/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":"2026-07-05T07:47:23Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"cR+etT9Km48IsvgpPsOpFByFbLaou/YOP4HhcPl0XtqOpVtHsoBYtDDGdW5/a3ULdPW67VvwlN5uPmU0hDd0Bw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-17T01:47:52.601665Z"},"content_sha256":"43f8dc6403ddcfa4e9ea7dfab4c7a6704a896f68844862e4bd0156ad797309b8","schema_version":"1.0","event_id":"sha256:43f8dc6403ddcfa4e9ea7dfab4c7a6704a896f68844862e4bd0156ad797309b8"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/EZHUFD3ZJX7ZAOEN4BMKSVNHR2/bundle.json","state_url":"https://pith.science/pith/EZHUFD3ZJX7ZAOEN4BMKSVNHR2/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/EZHUFD3ZJX7ZAOEN4BMKSVNHR2/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":"2026-08-17T01:47:52Z","links":{"resolver":"https://pith.science/pith/EZHUFD3ZJX7ZAOEN4BMKSVNHR2","bundle":"https://pith.science/pith/EZHUFD3ZJX7ZAOEN4BMKSVNHR2/bundle.json","state":"https://pith.science/pith/EZHUFD3ZJX7ZAOEN4BMKSVNHR2/state.json","well_known_bundle":"https://pith.science/.well-known/pith/EZHUFD3ZJX7ZAOEN4BMKSVNHR2/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2021:EZHUFD3ZJX7ZAOEN4BMKSVNHR2","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":"63ae58c9d996e91e42726851562e781429d169e8d37e9be58934729cc0d059ec","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-10-26T22:03:51Z","title_canon_sha256":"143861749e0631ad1931e3db718686390ab8a5b340606a75b7c30f9eb4a4b74a"},"schema_version":"1.0","source":{"id":"2110.14048","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2110.14048","created_at":"2026-07-05T07:47:23Z"},{"alias_kind":"arxiv_version","alias_value":"2110.14048v2","created_at":"2026-07-05T07:47:23Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2110.14048","created_at":"2026-07-05T07:47:23Z"},{"alias_kind":"pith_short_12","alias_value":"EZHUFD3ZJX7Z","created_at":"2026-07-05T07:47:23Z"},{"alias_kind":"pith_short_16","alias_value":"EZHUFD3ZJX7ZAOEN","created_at":"2026-07-05T07:47:23Z"},{"alias_kind":"pith_short_8","alias_value":"EZHUFD3Z","created_at":"2026-07-05T07:47:23Z"}],"graph_snapshots":[{"event_id":"sha256:43f8dc6403ddcfa4e9ea7dfab4c7a6704a896f68844862e4bd0156ad797309b8","target":"graph","created_at":"2026-07-05T07:47:23Z","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/2110.14048/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"The goal of multi-task learning is to enable more efficient learning than single task learning by sharing model structures for a diverse set of tasks. A standard multi-task learning objective is to minimize the average loss across all tasks. While straightforward, using this objective often results in much worse final performance for each task than learning them independently. A major challenge in optimizing a multi-task model is the conflicting gradients, where gradients of different task objectives are not well aligned so that following the average gradient direction can be detrimental to sp","authors_text":"Bo Liu, Peter Stone, Qiang Liu, Xiaojie Jin, Xingchao Liu","cross_cats":["cs.AI"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-10-26T22:03:51Z","title":"Conflict-Averse Gradient Descent for Multi-task Learning"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2110.14048","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:a5ecdaf494954019b180379a3bb5f2be531195f98d8a8fd00455a68e95c959d9","target":"record","created_at":"2026-07-05T07:47:23Z","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":"63ae58c9d996e91e42726851562e781429d169e8d37e9be58934729cc0d059ec","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-10-26T22:03:51Z","title_canon_sha256":"143861749e0631ad1931e3db718686390ab8a5b340606a75b7c30f9eb4a4b74a"},"schema_version":"1.0","source":{"id":"2110.14048","kind":"arxiv","version":2}},"canonical_sha256":"264f428f794dff90388de058a955a78e85f4348332cb1ceaa5f125aadb34d648","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"264f428f794dff90388de058a955a78e85f4348332cb1ceaa5f125aadb34d648","first_computed_at":"2026-07-05T07:47:23.361154Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T07:47:23.361154Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"BXcKl1Vuhm9JV6iirCGMbqSfET2Vqc+CXN01jmcFC3JnoGPibr0ME42+/WjSiPgONpMg9YhFwTBvajBsOVZ1Bg==","signature_status":"signed_v1","signed_at":"2026-07-05T07:47:23.361645Z","signed_message":"canonical_sha256_bytes"},"source_id":"2110.14048","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:a5ecdaf494954019b180379a3bb5f2be531195f98d8a8fd00455a68e95c959d9","sha256:43f8dc6403ddcfa4e9ea7dfab4c7a6704a896f68844862e4bd0156ad797309b8"],"state_sha256":"7156d6980c1497016e7c577a07f6d9f4a56deb9e9b2661e046ee2031d5ee66c9"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"DgvXbgmbL1jWb9gdMDp+EBPXSmCXr6RjbRHjRIeiKmWK3J70lYVHh1QTTwgzXmCN8rH1QJAi/McwW4bA9ysWDA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-17T01:47:52.607190Z","bundle_sha256":"a97fd6575eba09af04305bca047c301cb977861ee70ce69dc4a7faabc7bd189e"}}