{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:XDIPRS7AZBJH2IRFSCJEPE6I3D","short_pith_number":"pith:XDIPRS7A","canonical_record":{"source":{"id":"2401.10491","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-01-19T05:02:46Z","cross_cats_sorted":[],"title_canon_sha256":"4c97a2bdf75c1a4151331fc37d7e59d17bb0e18e10f8d5156f65faf3904a7379","abstract_canon_sha256":"c91771f898c304cbfddabdf5f686eb6a164bc731c3624a7b7d0938d756dabfe3"},"schema_version":"1.0"},"canonical_sha256":"b8d0f8cbe0c8527d222590924793c8d8e11ac15a9fafb85fbb869734e48d7888","source":{"kind":"arxiv","id":"2401.10491","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2401.10491","created_at":"2026-07-05T07:36:06Z"},{"alias_kind":"arxiv_version","alias_value":"2401.10491v2","created_at":"2026-07-05T07:36:06Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2401.10491","created_at":"2026-07-05T07:36:06Z"},{"alias_kind":"pith_short_12","alias_value":"XDIPRS7AZBJH","created_at":"2026-07-05T07:36:06Z"},{"alias_kind":"pith_short_16","alias_value":"XDIPRS7AZBJH2IRF","created_at":"2026-07-05T07:36:06Z"},{"alias_kind":"pith_short_8","alias_value":"XDIPRS7A","created_at":"2026-07-05T07:36:06Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:XDIPRS7AZBJH2IRFSCJEPE6I3D","target":"record","payload":{"canonical_record":{"source":{"id":"2401.10491","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-01-19T05:02:46Z","cross_cats_sorted":[],"title_canon_sha256":"4c97a2bdf75c1a4151331fc37d7e59d17bb0e18e10f8d5156f65faf3904a7379","abstract_canon_sha256":"c91771f898c304cbfddabdf5f686eb6a164bc731c3624a7b7d0938d756dabfe3"},"schema_version":"1.0"},"canonical_sha256":"b8d0f8cbe0c8527d222590924793c8d8e11ac15a9fafb85fbb869734e48d7888","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:36:06.429438Z","signature_b64":"mbBVl/eNvhUmBPmbnRI5goxeQzJUHeY/RowBEn1xXvcsLeHUWMcxBMhSw85++iShTzgPfVZG9AJ16wFlKe6QCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b8d0f8cbe0c8527d222590924793c8d8e11ac15a9fafb85fbb869734e48d7888","last_reissued_at":"2026-07-05T07:36:06.428940Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:36:06.428940Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2401.10491","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":"2026-07-05T07:36:06Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"E8zUGge/b0PNIRz3VsChsosT2VXcetqW23TCJ0vxf4yKbN8HUsd78TKkfTIqKNeizxEIND56xVjlZE+voe4eDA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-05T11:00:17.747869Z"},"content_sha256":"40d744737565c629e9f5b6b36e23270a26715356f4e64db6a884ae1d2389aa29","schema_version":"1.0","event_id":"sha256:40d744737565c629e9f5b6b36e23270a26715356f4e64db6a884ae1d2389aa29"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:XDIPRS7AZBJH2IRFSCJEPE6I3D","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Knowledge Fusion of Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Deng Cai, Fanqi Wan, Shuming Shi, Wei Bi, Xiaojun Quan, Xinting Huang","submitted_at":"2024-01-19T05:02:46Z","abstract_excerpt":"While training large language models (LLMs) from scratch can generate models with distinct functionalities and strengths, it comes at significant costs and may result in redundant capabilities. Alternatively, a cost-effective and compelling approach is to merge existing pre-trained LLMs into a more potent model. However, due to the varying architectures of these LLMs, directly blending their weights is impractical. In this paper, we introduce the notion of knowledge fusion for LLMs, aimed at combining the capabilities of existing LLMs and transferring them into a single LLM. By leveraging the "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2401.10491","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/2401.10491/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":"2026-07-05T07:36:06Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Q66a7w/CTL/TPLFC1QZRJGxK2x3CS68Kl4JymcwNdXLGSu0eLyjbMw37CtFSDKSfR4kaxAfd19iUckBp0JuNDQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-05T11:00:17.748399Z"},"content_sha256":"8a3094e2862f3f6b570cfb5dd3f57f0a9ed67863c966f2d9b4d5cc9a5d4981a9","schema_version":"1.0","event_id":"sha256:8a3094e2862f3f6b570cfb5dd3f57f0a9ed67863c966f2d9b4d5cc9a5d4981a9"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/XDIPRS7AZBJH2IRFSCJEPE6I3D/bundle.json","state_url":"https://pith.science/pith/XDIPRS7AZBJH2IRFSCJEPE6I3D/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/XDIPRS7AZBJH2IRFSCJEPE6I3D/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":"2026-08-05T11:00:17Z","links":{"resolver":"https://pith.science/pith/XDIPRS7AZBJH2IRFSCJEPE6I3D","bundle":"https://pith.science/pith/XDIPRS7AZBJH2IRFSCJEPE6I3D/bundle.json","state":"https://pith.science/pith/XDIPRS7AZBJH2IRFSCJEPE6I3D/state.json","well_known_bundle":"https://pith.science/.well-known/pith/XDIPRS7AZBJH2IRFSCJEPE6I3D/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:XDIPRS7AZBJH2IRFSCJEPE6I3D","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":"c91771f898c304cbfddabdf5f686eb6a164bc731c3624a7b7d0938d756dabfe3","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-01-19T05:02:46Z","title_canon_sha256":"4c97a2bdf75c1a4151331fc37d7e59d17bb0e18e10f8d5156f65faf3904a7379"},"schema_version":"1.0","source":{"id":"2401.10491","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2401.10491","created_at":"2026-07-05T07:36:06Z"},{"alias_kind":"arxiv_version","alias_value":"2401.10491v2","created_at":"2026-07-05T07:36:06Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2401.10491","created_at":"2026-07-05T07:36:06Z"},{"alias_kind":"pith_short_12","alias_value":"XDIPRS7AZBJH","created_at":"2026-07-05T07:36:06Z"},{"alias_kind":"pith_short_16","alias_value":"XDIPRS7AZBJH2IRF","created_at":"2026-07-05T07:36:06Z"},{"alias_kind":"pith_short_8","alias_value":"XDIPRS7A","created_at":"2026-07-05T07:36:06Z"}],"graph_snapshots":[{"event_id":"sha256:8a3094e2862f3f6b570cfb5dd3f57f0a9ed67863c966f2d9b4d5cc9a5d4981a9","target":"graph","created_at":"2026-07-05T07:36:06Z","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/2401.10491/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"While training large language models (LLMs) from scratch can generate models with distinct functionalities and strengths, it comes at significant costs and may result in redundant capabilities. Alternatively, a cost-effective and compelling approach is to merge existing pre-trained LLMs into a more potent model. However, due to the varying architectures of these LLMs, directly blending their weights is impractical. In this paper, we introduce the notion of knowledge fusion for LLMs, aimed at combining the capabilities of existing LLMs and transferring them into a single LLM. By leveraging the ","authors_text":"Deng Cai, Fanqi Wan, Shuming Shi, Wei Bi, Xiaojun Quan, Xinting Huang","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-01-19T05:02:46Z","title":"Knowledge Fusion of Large Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2401.10491","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":"sha256:40d744737565c629e9f5b6b36e23270a26715356f4e64db6a884ae1d2389aa29","target":"record","created_at":"2026-07-05T07:36:06Z","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":"c91771f898c304cbfddabdf5f686eb6a164bc731c3624a7b7d0938d756dabfe3","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-01-19T05:02:46Z","title_canon_sha256":"4c97a2bdf75c1a4151331fc37d7e59d17bb0e18e10f8d5156f65faf3904a7379"},"schema_version":"1.0","source":{"id":"2401.10491","kind":"arxiv","version":2}},"canonical_sha256":"b8d0f8cbe0c8527d222590924793c8d8e11ac15a9fafb85fbb869734e48d7888","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"b8d0f8cbe0c8527d222590924793c8d8e11ac15a9fafb85fbb869734e48d7888","first_computed_at":"2026-07-05T07:36:06.428940Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T07:36:06.428940Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"mbBVl/eNvhUmBPmbnRI5goxeQzJUHeY/RowBEn1xXvcsLeHUWMcxBMhSw85++iShTzgPfVZG9AJ16wFlKe6QCA==","signature_status":"signed_v1","signed_at":"2026-07-05T07:36:06.429438Z","signed_message":"canonical_sha256_bytes"},"source_id":"2401.10491","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:40d744737565c629e9f5b6b36e23270a26715356f4e64db6a884ae1d2389aa29","sha256:8a3094e2862f3f6b570cfb5dd3f57f0a9ed67863c966f2d9b4d5cc9a5d4981a9"],"state_sha256":"bbb5b304f21bd133060b68d15d10d00003caf9dbe2c3872638b12885a2f2f055"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"muIDL6A73cZp3ttnaVTp7gbERsHiiZ6U+mVKYyno5gx0R9ZhOWcBIUGqrYEOhQgLiNG9L6ghJHhxd/yastOSCg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-05T11:00:17.753024Z","bundle_sha256":"e4fff5c0e74f6d92e23b9714a3068206191695e46b9c4dd1f3ee25d464d6af5b"}}