{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:3HPBGM6U4J5NCDPO6GG7H4RJRW","short_pith_number":"pith:3HPBGM6U","canonical_record":{"source":{"id":"2605.26246","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-25T18:12:54Z","cross_cats_sorted":[],"title_canon_sha256":"6fe57c06310f22c648d929e1865ffa1f6f6f0d8b9c2dd009a734f4f1d112dbaf","abstract_canon_sha256":"1d08ea3b1485a0efebcb9039e1a9e3a8e3c08dbc3d68c5b63a3a17cccd0cb754"},"schema_version":"1.0"},"canonical_sha256":"d9de1333d4e27ad10deef18df3f2298dad809efec6cf6f031f3a8641d58651ff","source":{"kind":"arxiv","id":"2605.26246","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.26246","created_at":"2026-05-27T01:05:07Z"},{"alias_kind":"arxiv_version","alias_value":"2605.26246v1","created_at":"2026-05-27T01:05:07Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.26246","created_at":"2026-05-27T01:05:07Z"},{"alias_kind":"pith_short_12","alias_value":"3HPBGM6U4J5N","created_at":"2026-05-27T01:05:07Z"},{"alias_kind":"pith_short_16","alias_value":"3HPBGM6U4J5NCDPO","created_at":"2026-05-27T01:05:07Z"},{"alias_kind":"pith_short_8","alias_value":"3HPBGM6U","created_at":"2026-05-27T01:05:07Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:3HPBGM6U4J5NCDPO6GG7H4RJRW","target":"record","payload":{"canonical_record":{"source":{"id":"2605.26246","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-25T18:12:54Z","cross_cats_sorted":[],"title_canon_sha256":"6fe57c06310f22c648d929e1865ffa1f6f6f0d8b9c2dd009a734f4f1d112dbaf","abstract_canon_sha256":"1d08ea3b1485a0efebcb9039e1a9e3a8e3c08dbc3d68c5b63a3a17cccd0cb754"},"schema_version":"1.0"},"canonical_sha256":"d9de1333d4e27ad10deef18df3f2298dad809efec6cf6f031f3a8641d58651ff","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-27T01:05:07.957981Z","signature_b64":"IRF9qP0WkVDe4zFne4wzzKW6kaLLR/GLC4cITTchu4oo9B2LHOX15kiI7U9ntsPxWhwo7hKj2t9zwt2O4EntCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d9de1333d4e27ad10deef18df3f2298dad809efec6cf6f031f3a8641d58651ff","last_reissued_at":"2026-05-27T01:05:07.956912Z","signature_status":"signed_v1","first_computed_at":"2026-05-27T01:05:07.956912Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.26246","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":"2026-05-27T01:05:07Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"eVslkHj0+JYM5qIQ2nxsXZSfIVUa9dW20QQ1xMW/7eZtNcBUp/ZJJyXY5QfpA2KFHgGI8xSrdsdbNRe95JMcAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-30T14:45:22.198225Z"},"content_sha256":"d5edd18526cbc2071cd9924e733511dc409277392033611572478fcd256366f0","schema_version":"1.0","event_id":"sha256:d5edd18526cbc2071cd9924e733511dc409277392033611572478fcd256366f0"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:3HPBGM6U4J5NCDPO6GG7H4RJRW","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"The Bridge-Garden Dilemma in LLM Distillation: Why Mixing Hard and Soft Labels Works","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Guanghui Wang, Jin-Wen Wu, Kaiwen Lv Kacuila, Longtao Huang, Qianqian Xu, Qingming Huang, Zhiyong Yang, Zitai Wang","submitted_at":"2026-05-25T18:12:54Z","abstract_excerpt":"Knowledge distillation (KD) transfers knowledge from a large teacher model to a smaller student. In language modeling, the student is trained either on tokens sampled from the teacher (hard labels) or the teacher's full next-token distribution (soft labels). Despite soft labels appear strictly richer, we find that mixing hard and soft labels consistently yields better results. Crucially, we show that this gain cannot be explained by closer teacher matching during training. Instead, it comes from reduced exposure bias, the mismatch between training and inference distributions. To explain this p"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.26246","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/2605.26246/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":"2026-05-27T01:05:07Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"HT5WX0x2EyRgeWfOTjJONm0qq2ObHBcxD+u04/ZKYFELHfOFEXIbFwFaiUf2rNor6FpIRuFWqyhuIJT3oi3hDw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-30T14:45:22.198655Z"},"content_sha256":"619901fc271b62b7a20087f383798c0f9c49383f918c95b2828122fa65f23bf1","schema_version":"1.0","event_id":"sha256:619901fc271b62b7a20087f383798c0f9c49383f918c95b2828122fa65f23bf1"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/3HPBGM6U4J5NCDPO6GG7H4RJRW/bundle.json","state_url":"https://pith.science/pith/3HPBGM6U4J5NCDPO6GG7H4RJRW/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/3HPBGM6U4J5NCDPO6GG7H4RJRW/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":"2026-05-30T14:45:22Z","links":{"resolver":"https://pith.science/pith/3HPBGM6U4J5NCDPO6GG7H4RJRW","bundle":"https://pith.science/pith/3HPBGM6U4J5NCDPO6GG7H4RJRW/bundle.json","state":"https://pith.science/pith/3HPBGM6U4J5NCDPO6GG7H4RJRW/state.json","well_known_bundle":"https://pith.science/.well-known/pith/3HPBGM6U4J5NCDPO6GG7H4RJRW/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:3HPBGM6U4J5NCDPO6GG7H4RJRW","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":"1d08ea3b1485a0efebcb9039e1a9e3a8e3c08dbc3d68c5b63a3a17cccd0cb754","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-25T18:12:54Z","title_canon_sha256":"6fe57c06310f22c648d929e1865ffa1f6f6f0d8b9c2dd009a734f4f1d112dbaf"},"schema_version":"1.0","source":{"id":"2605.26246","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.26246","created_at":"2026-05-27T01:05:07Z"},{"alias_kind":"arxiv_version","alias_value":"2605.26246v1","created_at":"2026-05-27T01:05:07Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.26246","created_at":"2026-05-27T01:05:07Z"},{"alias_kind":"pith_short_12","alias_value":"3HPBGM6U4J5N","created_at":"2026-05-27T01:05:07Z"},{"alias_kind":"pith_short_16","alias_value":"3HPBGM6U4J5NCDPO","created_at":"2026-05-27T01:05:07Z"},{"alias_kind":"pith_short_8","alias_value":"3HPBGM6U","created_at":"2026-05-27T01:05:07Z"}],"graph_snapshots":[{"event_id":"sha256:619901fc271b62b7a20087f383798c0f9c49383f918c95b2828122fa65f23bf1","target":"graph","created_at":"2026-05-27T01:05:07Z","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/2605.26246/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Knowledge distillation (KD) transfers knowledge from a large teacher model to a smaller student. In language modeling, the student is trained either on tokens sampled from the teacher (hard labels) or the teacher's full next-token distribution (soft labels). Despite soft labels appear strictly richer, we find that mixing hard and soft labels consistently yields better results. Crucially, we show that this gain cannot be explained by closer teacher matching during training. Instead, it comes from reduced exposure bias, the mismatch between training and inference distributions. To explain this p","authors_text":"Guanghui Wang, Jin-Wen Wu, Kaiwen Lv Kacuila, Longtao Huang, Qianqian Xu, Qingming Huang, Zhiyong Yang, Zitai Wang","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-25T18:12:54Z","title":"The Bridge-Garden Dilemma in LLM Distillation: Why Mixing Hard and Soft Labels Works"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.26246","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:d5edd18526cbc2071cd9924e733511dc409277392033611572478fcd256366f0","target":"record","created_at":"2026-05-27T01:05:07Z","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":"1d08ea3b1485a0efebcb9039e1a9e3a8e3c08dbc3d68c5b63a3a17cccd0cb754","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-25T18:12:54Z","title_canon_sha256":"6fe57c06310f22c648d929e1865ffa1f6f6f0d8b9c2dd009a734f4f1d112dbaf"},"schema_version":"1.0","source":{"id":"2605.26246","kind":"arxiv","version":1}},"canonical_sha256":"d9de1333d4e27ad10deef18df3f2298dad809efec6cf6f031f3a8641d58651ff","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"d9de1333d4e27ad10deef18df3f2298dad809efec6cf6f031f3a8641d58651ff","first_computed_at":"2026-05-27T01:05:07.956912Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-27T01:05:07.956912Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"IRF9qP0WkVDe4zFne4wzzKW6kaLLR/GLC4cITTchu4oo9B2LHOX15kiI7U9ntsPxWhwo7hKj2t9zwt2O4EntCg==","signature_status":"signed_v1","signed_at":"2026-05-27T01:05:07.957981Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.26246","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:d5edd18526cbc2071cd9924e733511dc409277392033611572478fcd256366f0","sha256:619901fc271b62b7a20087f383798c0f9c49383f918c95b2828122fa65f23bf1"],"state_sha256":"9aa1776d998bc100f772e994e6e80c98132eae2994d948a24b37fa736936b770"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"YCVJyPbW+TMLZQSXrAtLcrNc/eYlGU2qE+gYOKt7k8ZensxlRegnRLaSRG0hMkEJBEQCLfk1bGtRhUPiQ6ShAQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-30T14:45:22.201773Z","bundle_sha256":"519133e71ddf8a48b334a2258cb99a1d24ebc769e2f927615a53141d88bf1076"}}