{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:EDPRMAQQNICDYZZWOC7ULZDMTQ","short_pith_number":"pith:EDPRMAQQ","canonical_record":{"source":{"id":"2507.01679","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-07-02T13:04:09Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"a52ab5ac2b944d0ba470efe078c004a67e7c87c49ef7dbabe131d33a6db54783","abstract_canon_sha256":"8390f738c802f7ba62a5207ad61243bb666cad5e61ba66c8b82c99959d80b745"},"schema_version":"1.0"},"canonical_sha256":"20df1602106a043c673670bf45e46c9c28ac9c128c239f55e486d5dd96db5230","source":{"kind":"arxiv","id":"2507.01679","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2507.01679","created_at":"2026-05-20T00:00:21Z"},{"alias_kind":"arxiv_version","alias_value":"2507.01679v3","created_at":"2026-05-20T00:00:21Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.01679","created_at":"2026-05-20T00:00:21Z"},{"alias_kind":"pith_short_12","alias_value":"EDPRMAQQNICD","created_at":"2026-05-20T00:00:21Z"},{"alias_kind":"pith_short_16","alias_value":"EDPRMAQQNICDYZZW","created_at":"2026-05-20T00:00:21Z"},{"alias_kind":"pith_short_8","alias_value":"EDPRMAQQ","created_at":"2026-05-20T00:00:21Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:EDPRMAQQNICDYZZWOC7ULZDMTQ","target":"record","payload":{"canonical_record":{"source":{"id":"2507.01679","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-07-02T13:04:09Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"a52ab5ac2b944d0ba470efe078c004a67e7c87c49ef7dbabe131d33a6db54783","abstract_canon_sha256":"8390f738c802f7ba62a5207ad61243bb666cad5e61ba66c8b82c99959d80b745"},"schema_version":"1.0"},"canonical_sha256":"20df1602106a043c673670bf45e46c9c28ac9c128c239f55e486d5dd96db5230","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:00:21.918166Z","signature_b64":"TtZjEXy2uVzNUb+RUuPP3HhQDkR0ph4F9RVsDbbx434IIN1BE4hAf3nGfci6QD654KTBnuoc3KjRT1ariMT2Bg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"20df1602106a043c673670bf45e46c9c28ac9c128c239f55e486d5dd96db5230","last_reissued_at":"2026-05-20T00:00:21.917416Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:00:21.917416Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2507.01679","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":"2026-05-20T00:00:21Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"UhG9U1bTzBd20wyWIRU4X5FBE99WKKPv99zGgAehe3VlVwob+6eFKfSYlnLryn47e3PMi2ViJf/WOfR7RTvHBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-25T20:08:07.240407Z"},"content_sha256":"29acee415dd34e60f3b71cf3a36d4ee2e7a90b0bdc36c4e0185e5c67c476bb53","schema_version":"1.0","event_id":"sha256:29acee415dd34e60f3b71cf3a36d4ee2e7a90b0bdc36c4e0185e5c67c476bb53"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:EDPRMAQQNICDYZZWOC7ULZDMTQ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Blending Supervised and Reinforcement Fine-Tuning with Prefix Sampling","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Edoardo M. Ponti, Ivan Titov, Tianhao Cheng, Yinghui Xu, Zeyu Huang, Zihan Qiu, Zili Wang","submitted_at":"2025-07-02T13:04:09Z","abstract_excerpt":"Existing LLMs-post-training techniques are broadly categorized into supervised fine-tuning (SFT) and reinforcement fine-tuning (RFT). Each paradigm presents a distinct trade-off: (1) SFT excels at mimicking demonstration data, but can lead to problematic generalization as a form of behavior cloning. (2) Conversely, RFT can significantly enhance a model's performance but is prone to learning unexpected behaviors, and its performance is sensitive to the initial policy. In this paper, we propose a unified view of these methods and introduce Prefix-RFT, a hybrid approach that synergizes learning f"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.01679","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/2507.01679/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":"2026-05-20T00:00:21Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"9dCQ7gzwVuXZ5+nmQ9T+pR4aLTrJNXtFu49MDMJ0yoVjuNw0uQBXinbSTD0NpSf/c2EGnbLKmv9SXrNWo+XbAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-25T20:08:07.241193Z"},"content_sha256":"304e767653b57e71bbb30121a6ceeda71781037e968000c54cc826824e6257a1","schema_version":"1.0","event_id":"sha256:304e767653b57e71bbb30121a6ceeda71781037e968000c54cc826824e6257a1"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/EDPRMAQQNICDYZZWOC7ULZDMTQ/bundle.json","state_url":"https://pith.science/pith/EDPRMAQQNICDYZZWOC7ULZDMTQ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/EDPRMAQQNICDYZZWOC7ULZDMTQ/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":"2026-05-25T20:08:07Z","links":{"resolver":"https://pith.science/pith/EDPRMAQQNICDYZZWOC7ULZDMTQ","bundle":"https://pith.science/pith/EDPRMAQQNICDYZZWOC7ULZDMTQ/bundle.json","state":"https://pith.science/pith/EDPRMAQQNICDYZZWOC7ULZDMTQ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/EDPRMAQQNICDYZZWOC7ULZDMTQ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:EDPRMAQQNICDYZZWOC7ULZDMTQ","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":"8390f738c802f7ba62a5207ad61243bb666cad5e61ba66c8b82c99959d80b745","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-07-02T13:04:09Z","title_canon_sha256":"a52ab5ac2b944d0ba470efe078c004a67e7c87c49ef7dbabe131d33a6db54783"},"schema_version":"1.0","source":{"id":"2507.01679","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2507.01679","created_at":"2026-05-20T00:00:21Z"},{"alias_kind":"arxiv_version","alias_value":"2507.01679v3","created_at":"2026-05-20T00:00:21Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.01679","created_at":"2026-05-20T00:00:21Z"},{"alias_kind":"pith_short_12","alias_value":"EDPRMAQQNICD","created_at":"2026-05-20T00:00:21Z"},{"alias_kind":"pith_short_16","alias_value":"EDPRMAQQNICDYZZW","created_at":"2026-05-20T00:00:21Z"},{"alias_kind":"pith_short_8","alias_value":"EDPRMAQQ","created_at":"2026-05-20T00:00:21Z"}],"graph_snapshots":[{"event_id":"sha256:304e767653b57e71bbb30121a6ceeda71781037e968000c54cc826824e6257a1","target":"graph","created_at":"2026-05-20T00:00:21Z","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/2507.01679/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Existing LLMs-post-training techniques are broadly categorized into supervised fine-tuning (SFT) and reinforcement fine-tuning (RFT). Each paradigm presents a distinct trade-off: (1) SFT excels at mimicking demonstration data, but can lead to problematic generalization as a form of behavior cloning. (2) Conversely, RFT can significantly enhance a model's performance but is prone to learning unexpected behaviors, and its performance is sensitive to the initial policy. In this paper, we propose a unified view of these methods and introduce Prefix-RFT, a hybrid approach that synergizes learning f","authors_text":"Edoardo M. Ponti, Ivan Titov, Tianhao Cheng, Yinghui Xu, Zeyu Huang, Zihan Qiu, Zili Wang","cross_cats":["cs.AI","cs.CL"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-07-02T13:04:09Z","title":"Blending Supervised and Reinforcement Fine-Tuning with Prefix Sampling"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.01679","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":"sha256:29acee415dd34e60f3b71cf3a36d4ee2e7a90b0bdc36c4e0185e5c67c476bb53","target":"record","created_at":"2026-05-20T00:00:21Z","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":"8390f738c802f7ba62a5207ad61243bb666cad5e61ba66c8b82c99959d80b745","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-07-02T13:04:09Z","title_canon_sha256":"a52ab5ac2b944d0ba470efe078c004a67e7c87c49ef7dbabe131d33a6db54783"},"schema_version":"1.0","source":{"id":"2507.01679","kind":"arxiv","version":3}},"canonical_sha256":"20df1602106a043c673670bf45e46c9c28ac9c128c239f55e486d5dd96db5230","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"20df1602106a043c673670bf45e46c9c28ac9c128c239f55e486d5dd96db5230","first_computed_at":"2026-05-20T00:00:21.917416Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-20T00:00:21.917416Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"TtZjEXy2uVzNUb+RUuPP3HhQDkR0ph4F9RVsDbbx434IIN1BE4hAf3nGfci6QD654KTBnuoc3KjRT1ariMT2Bg==","signature_status":"signed_v1","signed_at":"2026-05-20T00:00:21.918166Z","signed_message":"canonical_sha256_bytes"},"source_id":"2507.01679","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:29acee415dd34e60f3b71cf3a36d4ee2e7a90b0bdc36c4e0185e5c67c476bb53","sha256:304e767653b57e71bbb30121a6ceeda71781037e968000c54cc826824e6257a1"],"state_sha256":"febb7b6e3231c93ba89c9678731541f0720f9d146a86d4d25d61ff674a0b2cf8"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"lMdeteutbsd8AgrcLdAr3GtyKHnfN05ODIE8EFfvq/rjzyTmwxWE2sN12Lj6oBNQZi2RMyd8x8sJKCfX+8WkCg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-25T20:08:07.245214Z","bundle_sha256":"141612a70f0c55d29d5469ff04d3283b3b74533cba63f4a29bced6c83ad7584d"}}