{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:4G2ZJFDGWQUK6S77IPWJMG7D2X","short_pith_number":"pith:4G2ZJFDG","canonical_record":{"source":{"id":"2504.18580","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-04-23T05:11:21Z","cross_cats_sorted":[],"title_canon_sha256":"630723b981950e106cafb978efb78114c0d92efd8027a9ef8834fd1215804abd","abstract_canon_sha256":"b222d72feb6be02c74b735f4fb44b1b360b4ca07255e132a59b0563a54a54566"},"schema_version":"1.0"},"canonical_sha256":"e1b5949466b428af4bff43ec961be3d5f142f7ca5581b7ed2b86b2704811a285","source":{"kind":"arxiv","id":"2504.18580","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2504.18580","created_at":"2026-07-05T10:54:17Z"},{"alias_kind":"arxiv_version","alias_value":"2504.18580v1","created_at":"2026-07-05T10:54:17Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.18580","created_at":"2026-07-05T10:54:17Z"},{"alias_kind":"pith_short_12","alias_value":"4G2ZJFDGWQUK","created_at":"2026-07-05T10:54:17Z"},{"alias_kind":"pith_short_16","alias_value":"4G2ZJFDGWQUK6S77","created_at":"2026-07-05T10:54:17Z"},{"alias_kind":"pith_short_8","alias_value":"4G2ZJFDG","created_at":"2026-07-05T10:54:17Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:4G2ZJFDGWQUK6S77IPWJMG7D2X","target":"record","payload":{"canonical_record":{"source":{"id":"2504.18580","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-04-23T05:11:21Z","cross_cats_sorted":[],"title_canon_sha256":"630723b981950e106cafb978efb78114c0d92efd8027a9ef8834fd1215804abd","abstract_canon_sha256":"b222d72feb6be02c74b735f4fb44b1b360b4ca07255e132a59b0563a54a54566"},"schema_version":"1.0"},"canonical_sha256":"e1b5949466b428af4bff43ec961be3d5f142f7ca5581b7ed2b86b2704811a285","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:54:17.467723Z","signature_b64":"joIBIbOBd2t4sEeCGlEhFm4SQa0GygCR/aJubXcen7nE4duaOiB1jrZmhoESFkmv4TWmrWWYw5Efe/i613d0BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e1b5949466b428af4bff43ec961be3d5f142f7ca5581b7ed2b86b2704811a285","last_reissued_at":"2026-07-05T10:54:17.467210Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:54:17.467210Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2504.18580","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:54:17Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"CbdszqPsDEZXV4Xagb2syCimg7yxWYgXtA57nL86nj8f0svdPvqaHwBGExfG0WlKCFb04nte2Mpy0mXW3RVTCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-17T12:35:26.520251Z"},"content_sha256":"20cab648543bc2513458429fee0e2f9f87465fb74e2ba3aab9cfcff53341dc2c","schema_version":"1.0","event_id":"sha256:20cab648543bc2513458429fee0e2f9f87465fb74e2ba3aab9cfcff53341dc2c"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:4G2ZJFDGWQUK6S77IPWJMG7D2X","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Parameter-Efficient Checkpoint Merging via Metrics-Weighted Averaging","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Sehyun Choi, Shi Jie Yu","submitted_at":"2025-04-23T05:11:21Z","abstract_excerpt":"Checkpoint merging is a technique for combining multiple model snapshots into a single superior model, potentially reducing training time for large language models. This paper explores checkpoint merging in the context of parameter-efficient fine-tuning (PEFT), where only small adapter modules (e.g. LoRA) are trained. We propose Metrics-Weighted Averaging (MWA), a simple yet effective method to merge model checkpoints by weighting their parameters according to performance metrics. In particular, we investigate weighting by training loss and by training steps, under the intuition that lower-los"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.18580","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/2504.18580/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":"2026-07-05T10:54:17Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"H7Her0IZgVfxFe2VUm6uSJcNOMX/DB8xxHOST5BLehprjFnJRoSYA7FZZqLzUBPk+63iqiBw0HaWrwUpnnkoDA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-17T12:35:26.520685Z"},"content_sha256":"e6d24c39de8956730ecc37c1176dc5aa04724a427fb460027a3213512b12f73c","schema_version":"1.0","event_id":"sha256:e6d24c39de8956730ecc37c1176dc5aa04724a427fb460027a3213512b12f73c"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/4G2ZJFDGWQUK6S77IPWJMG7D2X/bundle.json","state_url":"https://pith.science/pith/4G2ZJFDGWQUK6S77IPWJMG7D2X/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/4G2ZJFDGWQUK6S77IPWJMG7D2X/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":"2026-08-17T12:35:26Z","links":{"resolver":"https://pith.science/pith/4G2ZJFDGWQUK6S77IPWJMG7D2X","bundle":"https://pith.science/pith/4G2ZJFDGWQUK6S77IPWJMG7D2X/bundle.json","state":"https://pith.science/pith/4G2ZJFDGWQUK6S77IPWJMG7D2X/state.json","well_known_bundle":"https://pith.science/.well-known/pith/4G2ZJFDGWQUK6S77IPWJMG7D2X/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:4G2ZJFDGWQUK6S77IPWJMG7D2X","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":"b222d72feb6be02c74b735f4fb44b1b360b4ca07255e132a59b0563a54a54566","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-04-23T05:11:21Z","title_canon_sha256":"630723b981950e106cafb978efb78114c0d92efd8027a9ef8834fd1215804abd"},"schema_version":"1.0","source":{"id":"2504.18580","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2504.18580","created_at":"2026-07-05T10:54:17Z"},{"alias_kind":"arxiv_version","alias_value":"2504.18580v1","created_at":"2026-07-05T10:54:17Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.18580","created_at":"2026-07-05T10:54:17Z"},{"alias_kind":"pith_short_12","alias_value":"4G2ZJFDGWQUK","created_at":"2026-07-05T10:54:17Z"},{"alias_kind":"pith_short_16","alias_value":"4G2ZJFDGWQUK6S77","created_at":"2026-07-05T10:54:17Z"},{"alias_kind":"pith_short_8","alias_value":"4G2ZJFDG","created_at":"2026-07-05T10:54:17Z"}],"graph_snapshots":[{"event_id":"sha256:e6d24c39de8956730ecc37c1176dc5aa04724a427fb460027a3213512b12f73c","target":"graph","created_at":"2026-07-05T10:54:17Z","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/2504.18580/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Checkpoint merging is a technique for combining multiple model snapshots into a single superior model, potentially reducing training time for large language models. This paper explores checkpoint merging in the context of parameter-efficient fine-tuning (PEFT), where only small adapter modules (e.g. LoRA) are trained. We propose Metrics-Weighted Averaging (MWA), a simple yet effective method to merge model checkpoints by weighting their parameters according to performance metrics. In particular, we investigate weighting by training loss and by training steps, under the intuition that lower-los","authors_text":"Sehyun Choi, Shi Jie Yu","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-04-23T05:11:21Z","title":"Parameter-Efficient Checkpoint Merging via Metrics-Weighted Averaging"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.18580","kind":"arxiv","version":1},"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":"sha256:20cab648543bc2513458429fee0e2f9f87465fb74e2ba3aab9cfcff53341dc2c","target":"record","created_at":"2026-07-05T10:54:17Z","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":"b222d72feb6be02c74b735f4fb44b1b360b4ca07255e132a59b0563a54a54566","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-04-23T05:11:21Z","title_canon_sha256":"630723b981950e106cafb978efb78114c0d92efd8027a9ef8834fd1215804abd"},"schema_version":"1.0","source":{"id":"2504.18580","kind":"arxiv","version":1}},"canonical_sha256":"e1b5949466b428af4bff43ec961be3d5f142f7ca5581b7ed2b86b2704811a285","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"e1b5949466b428af4bff43ec961be3d5f142f7ca5581b7ed2b86b2704811a285","first_computed_at":"2026-07-05T10:54:17.467210Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:54:17.467210Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"joIBIbOBd2t4sEeCGlEhFm4SQa0GygCR/aJubXcen7nE4duaOiB1jrZmhoESFkmv4TWmrWWYw5Efe/i613d0BQ==","signature_status":"signed_v1","signed_at":"2026-07-05T10:54:17.467723Z","signed_message":"canonical_sha256_bytes"},"source_id":"2504.18580","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:20cab648543bc2513458429fee0e2f9f87465fb74e2ba3aab9cfcff53341dc2c","sha256:e6d24c39de8956730ecc37c1176dc5aa04724a427fb460027a3213512b12f73c"],"state_sha256":"9263a9493cd947cf85c0f8e3b9774a6aa349567676f36925b15178130f8668ee"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Di4p8Y2J4VK/rFpvhq1+GZ94NQe9gVZVB1dwzKHQTEQZYRRJ1C4E7x98Qz1RZuV61cYTdMxaYmC7WK7eEjF2CQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-17T12:35:26.523158Z","bundle_sha256":"72c15e2156c4e66289655a7b00e627c45cd33f6b1353915cf6716675316a59a2"}}