{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:27CJUL25ICHTODOTPYRRYEO3KH","short_pith_number":"pith:27CJUL25","canonical_record":{"source":{"id":"2509.02108","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-09-02T09:04:41Z","cross_cats_sorted":[],"title_canon_sha256":"ac949e8702ab4e938a06392afc88419baffc945112bcee9149a77c3a24e3c459","abstract_canon_sha256":"f1abc0599ce26574a108e7991257b702d949a19f458b5b2580a97863074b6239"},"schema_version":"1.0"},"canonical_sha256":"d7c49a2f5d408f370dd37e231c11db51ccd8ac8dd25cad4f64812b37a006e808","source":{"kind":"arxiv","id":"2509.02108","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2509.02108","created_at":"2026-07-05T12:09:46Z"},{"alias_kind":"arxiv_version","alias_value":"2509.02108v3","created_at":"2026-07-05T12:09:46Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.02108","created_at":"2026-07-05T12:09:46Z"},{"alias_kind":"pith_short_12","alias_value":"27CJUL25ICHT","created_at":"2026-07-05T12:09:46Z"},{"alias_kind":"pith_short_16","alias_value":"27CJUL25ICHTODOT","created_at":"2026-07-05T12:09:46Z"},{"alias_kind":"pith_short_8","alias_value":"27CJUL25","created_at":"2026-07-05T12:09:46Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:27CJUL25ICHTODOTPYRRYEO3KH","target":"record","payload":{"canonical_record":{"source":{"id":"2509.02108","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-09-02T09:04:41Z","cross_cats_sorted":[],"title_canon_sha256":"ac949e8702ab4e938a06392afc88419baffc945112bcee9149a77c3a24e3c459","abstract_canon_sha256":"f1abc0599ce26574a108e7991257b702d949a19f458b5b2580a97863074b6239"},"schema_version":"1.0"},"canonical_sha256":"d7c49a2f5d408f370dd37e231c11db51ccd8ac8dd25cad4f64812b37a006e808","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:09:46.145194Z","signature_b64":"syTYfzdwndgCO2OS+IRYmDWAuxWJ6yY1IHyPgUIavIIPw1UfgaBKJYk20OP9P8ExWYQ/VzjtfQAxJpFXN6O6Aw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d7c49a2f5d408f370dd37e231c11db51ccd8ac8dd25cad4f64812b37a006e808","last_reissued_at":"2026-07-05T12:09:46.144721Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:09:46.144721Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2509.02108","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T12:09:46Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"6doxYXgI7/6wzSpCqXzpzZmVgqoI1c8MQvdpMVv7LPb1OmMo6Yr4GBRPv2g3nGXW68rjVDxt+tHj3esuJ35JAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-06T13:45:44.083407Z"},"content_sha256":"6cbf48b755a7f5286e5fb2c8601aa5a355be17baa1083d32ad8adc440ec4b5dc","schema_version":"1.0","event_id":"sha256:6cbf48b755a7f5286e5fb2c8601aa5a355be17baa1083d32ad8adc440ec4b5dc"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:27CJUL25ICHTODOTPYRRYEO3KH","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"DivMerge: A divergence-based model merging method for multi-tasking","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Brahim Touayouch, G\\'eraldine Damnati, Gw\\'enol\\'e Lecorv\\'e, Lo\\\"ic Fosse","submitted_at":"2025-09-02T09:04:41Z","abstract_excerpt":"Multi-task learning (MTL) is often achieved by merging datasets before fine-tuning, but the growing availability of fine-tuned models has led to new approaches such as model merging via task arithmetic. A major challenge in this setting is task interference, which worsens as the number of tasks increases. We propose a method that merges models trained on different tasks into a single model, maintaining strong performance across all tasks. Our approach leverages Jensen-Shannon divergence to guide the merging process without requiring additional labelled data, and automatically balances task imp"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.02108","kind":"arxiv","version":3},"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/2509.02108/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":"2026-07-05T12:09:46Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"+HETN2KfO9K0qrQ+RDWFjPIVwDplwpAZsAeksD6QS0XioNMk6DhJjActt+ijknU+HQiA+BxXTvJaml7TXSKKCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-06T13:45:44.083783Z"},"content_sha256":"6b6991e695997dc21c4d6890d3d7575d2b33220dbe2d007e3a6e4350b4a2d7dc","schema_version":"1.0","event_id":"sha256:6b6991e695997dc21c4d6890d3d7575d2b33220dbe2d007e3a6e4350b4a2d7dc"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/27CJUL25ICHTODOTPYRRYEO3KH/bundle.json","state_url":"https://pith.science/pith/27CJUL25ICHTODOTPYRRYEO3KH/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/27CJUL25ICHTODOTPYRRYEO3KH/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":"2026-07-06T13:45:44Z","links":{"resolver":"https://pith.science/pith/27CJUL25ICHTODOTPYRRYEO3KH","bundle":"https://pith.science/pith/27CJUL25ICHTODOTPYRRYEO3KH/bundle.json","state":"https://pith.science/pith/27CJUL25ICHTODOTPYRRYEO3KH/state.json","well_known_bundle":"https://pith.science/.well-known/pith/27CJUL25ICHTODOTPYRRYEO3KH/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:27CJUL25ICHTODOTPYRRYEO3KH","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":"f1abc0599ce26574a108e7991257b702d949a19f458b5b2580a97863074b6239","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-09-02T09:04:41Z","title_canon_sha256":"ac949e8702ab4e938a06392afc88419baffc945112bcee9149a77c3a24e3c459"},"schema_version":"1.0","source":{"id":"2509.02108","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2509.02108","created_at":"2026-07-05T12:09:46Z"},{"alias_kind":"arxiv_version","alias_value":"2509.02108v3","created_at":"2026-07-05T12:09:46Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.02108","created_at":"2026-07-05T12:09:46Z"},{"alias_kind":"pith_short_12","alias_value":"27CJUL25ICHT","created_at":"2026-07-05T12:09:46Z"},{"alias_kind":"pith_short_16","alias_value":"27CJUL25ICHTODOT","created_at":"2026-07-05T12:09:46Z"},{"alias_kind":"pith_short_8","alias_value":"27CJUL25","created_at":"2026-07-05T12:09:46Z"}],"graph_snapshots":[{"event_id":"sha256:6b6991e695997dc21c4d6890d3d7575d2b33220dbe2d007e3a6e4350b4a2d7dc","target":"graph","created_at":"2026-07-05T12:09:46Z","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/2509.02108/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Multi-task learning (MTL) is often achieved by merging datasets before fine-tuning, but the growing availability of fine-tuned models has led to new approaches such as model merging via task arithmetic. A major challenge in this setting is task interference, which worsens as the number of tasks increases. We propose a method that merges models trained on different tasks into a single model, maintaining strong performance across all tasks. Our approach leverages Jensen-Shannon divergence to guide the merging process without requiring additional labelled data, and automatically balances task imp","authors_text":"Brahim Touayouch, G\\'eraldine Damnati, Gw\\'enol\\'e Lecorv\\'e, Lo\\\"ic Fosse","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-09-02T09:04:41Z","title":"DivMerge: A divergence-based model merging method for multi-tasking"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.02108","kind":"arxiv","version":3},"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":"sha256:6cbf48b755a7f5286e5fb2c8601aa5a355be17baa1083d32ad8adc440ec4b5dc","target":"record","created_at":"2026-07-05T12:09:46Z","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":"f1abc0599ce26574a108e7991257b702d949a19f458b5b2580a97863074b6239","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-09-02T09:04:41Z","title_canon_sha256":"ac949e8702ab4e938a06392afc88419baffc945112bcee9149a77c3a24e3c459"},"schema_version":"1.0","source":{"id":"2509.02108","kind":"arxiv","version":3}},"canonical_sha256":"d7c49a2f5d408f370dd37e231c11db51ccd8ac8dd25cad4f64812b37a006e808","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"d7c49a2f5d408f370dd37e231c11db51ccd8ac8dd25cad4f64812b37a006e808","first_computed_at":"2026-07-05T12:09:46.144721Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T12:09:46.144721Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"syTYfzdwndgCO2OS+IRYmDWAuxWJ6yY1IHyPgUIavIIPw1UfgaBKJYk20OP9P8ExWYQ/VzjtfQAxJpFXN6O6Aw==","signature_status":"signed_v1","signed_at":"2026-07-05T12:09:46.145194Z","signed_message":"canonical_sha256_bytes"},"source_id":"2509.02108","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:6cbf48b755a7f5286e5fb2c8601aa5a355be17baa1083d32ad8adc440ec4b5dc","sha256:6b6991e695997dc21c4d6890d3d7575d2b33220dbe2d007e3a6e4350b4a2d7dc"],"state_sha256":"0c14d6b5476894476f1b0b7b1ecd5558b072c525438b2665de1fc12f8d8a5d81"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"eV9xCLkNlWznZnh5b308fRL6yIH6rK+re8Ugfvdtp9penc8JvqpSTJMXJ/zVJ/KiUuuk8CCQLaUyL7UV4Sn0AA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-06T13:45:44.086084Z","bundle_sha256":"c8c0eb84d2ded72eca40bc8b397e6c7f17581d17d07fea8b8863209f6e207d76"}}