{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:JIHRMEGIP63V2OQ4Q4YXLMAZID","short_pith_number":"pith:JIHRMEGI","canonical_record":{"source":{"id":"2505.08507","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-05-13T12:37:48Z","cross_cats_sorted":[],"title_canon_sha256":"c62c8c07852bbff96004bf0e94df8b6fffe3ef1336aaec46bc29edeacf90d834","abstract_canon_sha256":"3f003d355eb3a7ff1222aebf1d8007b65cda1627e6114cb86456b114905e62c0"},"schema_version":"1.0"},"canonical_sha256":"4a0f1610c87fb75d3a1c873175b01940fc43a0433e649644c8079c9729ae34d0","source":{"kind":"arxiv","id":"2505.08507","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2505.08507","created_at":"2026-07-05T11:02:31Z"},{"alias_kind":"arxiv_version","alias_value":"2505.08507v1","created_at":"2026-07-05T11:02:31Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.08507","created_at":"2026-07-05T11:02:31Z"},{"alias_kind":"pith_short_12","alias_value":"JIHRMEGIP63V","created_at":"2026-07-05T11:02:31Z"},{"alias_kind":"pith_short_16","alias_value":"JIHRMEGIP63V2OQ4","created_at":"2026-07-05T11:02:31Z"},{"alias_kind":"pith_short_8","alias_value":"JIHRMEGI","created_at":"2026-07-05T11:02:31Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:JIHRMEGIP63V2OQ4Q4YXLMAZID","target":"record","payload":{"canonical_record":{"source":{"id":"2505.08507","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-05-13T12:37:48Z","cross_cats_sorted":[],"title_canon_sha256":"c62c8c07852bbff96004bf0e94df8b6fffe3ef1336aaec46bc29edeacf90d834","abstract_canon_sha256":"3f003d355eb3a7ff1222aebf1d8007b65cda1627e6114cb86456b114905e62c0"},"schema_version":"1.0"},"canonical_sha256":"4a0f1610c87fb75d3a1c873175b01940fc43a0433e649644c8079c9729ae34d0","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:02:31.528134Z","signature_b64":"2/dfsMtdLX0RH5YppNuMEFBK1C1KRSGCQbAx0LvBBpQMaHIdpfYgvKp34K560LY+8EJPjkz7EZbRRslVUJ/PDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4a0f1610c87fb75d3a1c873175b01940fc43a0433e649644c8079c9729ae34d0","last_reissued_at":"2026-07-05T11:02:31.527507Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:02:31.527507Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2505.08507","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":"2026-07-05T11:02:31Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"XTxM59FNHrl+AaypT6OiYpGujtFYKsD/1GIWpkh575UkOTe5NyH9oLHKdx4i6mp1GgznXYVglo/2wESyflg/AA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-07T12:45:20.861626Z"},"content_sha256":"563c274830f0957702135dd8d5e65db5b9078fd082f0af24f505a87b89216f1d","schema_version":"1.0","event_id":"sha256:563c274830f0957702135dd8d5e65db5b9078fd082f0af24f505a87b89216f1d"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:JIHRMEGIP63V2OQ4Q4YXLMAZID","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"InfoPO: On Mutual Information Maximization for Large Language Model Alignment","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Julian Katz-Samuels, Marc Versage, Qingjun Cui, Sujay Sanghavi, Teng Xiao, Tian Wang, Trishul Chilimbi, Zhen Ge","submitted_at":"2025-05-13T12:37:48Z","abstract_excerpt":"We study the post-training of large language models (LLMs) with human preference data. Recently, direct preference optimization and its variants have shown considerable promise in aligning language models, eliminating the need for reward models and online sampling. Despite these benefits, these methods rely on explicit assumptions about the Bradley-Terry (BT) model, which makes them prone to overfitting and results in suboptimal performance, particularly on reasoning-heavy tasks. To address these challenges, we propose a principled preference fine-tuning algorithm called InfoPO, which effectiv"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.08507","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/2505.08507/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":"2026-07-05T11:02:31Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"1mtc8hfdmhWDfWZ8bnPuDc+KrSkuHC05wrwFVyBJN1zZAuqJgiShxlZotvF/g0ufBzhB0BZWSyKhmbtBQOEEAg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-07T12:45:20.861998Z"},"content_sha256":"a5eba1d93877c5ecd00b4fab8308408bf8b23b34bcf0071cc761d2e6755525a1","schema_version":"1.0","event_id":"sha256:a5eba1d93877c5ecd00b4fab8308408bf8b23b34bcf0071cc761d2e6755525a1"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/JIHRMEGIP63V2OQ4Q4YXLMAZID/bundle.json","state_url":"https://pith.science/pith/JIHRMEGIP63V2OQ4Q4YXLMAZID/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/JIHRMEGIP63V2OQ4Q4YXLMAZID/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":"2026-07-07T12:45:20Z","links":{"resolver":"https://pith.science/pith/JIHRMEGIP63V2OQ4Q4YXLMAZID","bundle":"https://pith.science/pith/JIHRMEGIP63V2OQ4Q4YXLMAZID/bundle.json","state":"https://pith.science/pith/JIHRMEGIP63V2OQ4Q4YXLMAZID/state.json","well_known_bundle":"https://pith.science/.well-known/pith/JIHRMEGIP63V2OQ4Q4YXLMAZID/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:JIHRMEGIP63V2OQ4Q4YXLMAZID","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":"3f003d355eb3a7ff1222aebf1d8007b65cda1627e6114cb86456b114905e62c0","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-05-13T12:37:48Z","title_canon_sha256":"c62c8c07852bbff96004bf0e94df8b6fffe3ef1336aaec46bc29edeacf90d834"},"schema_version":"1.0","source":{"id":"2505.08507","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2505.08507","created_at":"2026-07-05T11:02:31Z"},{"alias_kind":"arxiv_version","alias_value":"2505.08507v1","created_at":"2026-07-05T11:02:31Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.08507","created_at":"2026-07-05T11:02:31Z"},{"alias_kind":"pith_short_12","alias_value":"JIHRMEGIP63V","created_at":"2026-07-05T11:02:31Z"},{"alias_kind":"pith_short_16","alias_value":"JIHRMEGIP63V2OQ4","created_at":"2026-07-05T11:02:31Z"},{"alias_kind":"pith_short_8","alias_value":"JIHRMEGI","created_at":"2026-07-05T11:02:31Z"}],"graph_snapshots":[{"event_id":"sha256:a5eba1d93877c5ecd00b4fab8308408bf8b23b34bcf0071cc761d2e6755525a1","target":"graph","created_at":"2026-07-05T11:02:31Z","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/2505.08507/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"We study the post-training of large language models (LLMs) with human preference data. Recently, direct preference optimization and its variants have shown considerable promise in aligning language models, eliminating the need for reward models and online sampling. Despite these benefits, these methods rely on explicit assumptions about the Bradley-Terry (BT) model, which makes them prone to overfitting and results in suboptimal performance, particularly on reasoning-heavy tasks. To address these challenges, we propose a principled preference fine-tuning algorithm called InfoPO, which effectiv","authors_text":"Julian Katz-Samuels, Marc Versage, Qingjun Cui, Sujay Sanghavi, Teng Xiao, Tian Wang, Trishul Chilimbi, Zhen Ge","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-05-13T12:37:48Z","title":"InfoPO: On Mutual Information Maximization for Large Language Model Alignment"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.08507","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":"sha256:563c274830f0957702135dd8d5e65db5b9078fd082f0af24f505a87b89216f1d","target":"record","created_at":"2026-07-05T11:02:31Z","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":"3f003d355eb3a7ff1222aebf1d8007b65cda1627e6114cb86456b114905e62c0","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-05-13T12:37:48Z","title_canon_sha256":"c62c8c07852bbff96004bf0e94df8b6fffe3ef1336aaec46bc29edeacf90d834"},"schema_version":"1.0","source":{"id":"2505.08507","kind":"arxiv","version":1}},"canonical_sha256":"4a0f1610c87fb75d3a1c873175b01940fc43a0433e649644c8079c9729ae34d0","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"4a0f1610c87fb75d3a1c873175b01940fc43a0433e649644c8079c9729ae34d0","first_computed_at":"2026-07-05T11:02:31.527507Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:02:31.527507Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"2/dfsMtdLX0RH5YppNuMEFBK1C1KRSGCQbAx0LvBBpQMaHIdpfYgvKp34K560LY+8EJPjkz7EZbRRslVUJ/PDA==","signature_status":"signed_v1","signed_at":"2026-07-05T11:02:31.528134Z","signed_message":"canonical_sha256_bytes"},"source_id":"2505.08507","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:563c274830f0957702135dd8d5e65db5b9078fd082f0af24f505a87b89216f1d","sha256:a5eba1d93877c5ecd00b4fab8308408bf8b23b34bcf0071cc761d2e6755525a1"],"state_sha256":"a03acea1725cc8f6b20db5b9444d248a15ea61f4f26d31043018ad389d5ea1ef"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"xr9/uUYaB3sfULiTGlC8P7Hv3e6ud4e8GELBP5eK9jhUzJpTg+5RLEn85FvC9u9U2ts18fwOCmSs8Tfo4HjABw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-07T12:45:20.863882Z","bundle_sha256":"8db5ddaea172717991d67c0c765e4f3951a1b51df3e1cd8ffbc5f01c78657eb7"}}