{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:L3RRNG4MJN47ZWVIMDGFLNWFHE","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":"75c520788dc9770af57b4650b026cf066d0fa512c5abd6fd4290f6b1f8eb169b","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-03-05T12:49:48Z","title_canon_sha256":"4be182aa8d8084fe071c1c8fe4e09754960b475e72ba3b50587f90967e261005"},"schema_version":"1.0","source":{"id":"2503.03460","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2503.03460","created_at":"2026-07-05T11:41:39Z"},{"alias_kind":"arxiv_version","alias_value":"2503.03460v2","created_at":"2026-07-05T11:41:39Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.03460","created_at":"2026-07-05T11:41:39Z"},{"alias_kind":"pith_short_12","alias_value":"L3RRNG4MJN47","created_at":"2026-07-05T11:41:39Z"},{"alias_kind":"pith_short_16","alias_value":"L3RRNG4MJN47ZWVI","created_at":"2026-07-05T11:41:39Z"},{"alias_kind":"pith_short_8","alias_value":"L3RRNG4M","created_at":"2026-07-05T11:41:39Z"}],"graph_snapshots":[{"event_id":"sha256:23ba634c2667ada4c57077ed4c9751855b5f1d9837162b5394a91626bb5034f3","target":"graph","created_at":"2026-07-05T11:41:39Z","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/2503.03460/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Fine-tuning Large Language Models (LLMs) with first-order methods like back-propagation is computationally intensive. Zeroth-Order (ZO) optimisation uses function evaluations instead of gradients, reducing memory usage, but suffers from slow convergence in high-dimensional models. As a result, ZO research in LLMs has mostly focused on classification, overlooking more complex generative tasks. In this paper, we introduce ZOPrO, a novel ZO algorithm designed for Preference Optimisation in LLMs. We begin by analysing the interplay between policy and reward models during traditional (first-order) ","authors_text":"Alessio Galatolo, Katie Winkle, Meriem Beloucif, Zhenbang Dai","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-03-05T12:49:48Z","title":"Visualising Policy-Reward Interplay to Inform Zeroth-Order Preference Optimisation of Large Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.03460","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":"sha256:98df2fab7ac25a7011c89b98290ee2ab1cb2ae6d138271eeb7dfc1c98542d034","target":"record","created_at":"2026-07-05T11:41:39Z","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":"75c520788dc9770af57b4650b026cf066d0fa512c5abd6fd4290f6b1f8eb169b","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-03-05T12:49:48Z","title_canon_sha256":"4be182aa8d8084fe071c1c8fe4e09754960b475e72ba3b50587f90967e261005"},"schema_version":"1.0","source":{"id":"2503.03460","kind":"arxiv","version":2}},"canonical_sha256":"5ee3169b8c4b79fcdaa860cc55b6c53906ca9d66b77900d37c9e6c27e9c21bf1","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"5ee3169b8c4b79fcdaa860cc55b6c53906ca9d66b77900d37c9e6c27e9c21bf1","first_computed_at":"2026-07-05T11:41:39.873028Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:41:39.873028Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"Qes3VyKC9gE+3e8cbpXXLZAHUNqJcMJMpP4/gbxqutygBTMT5FLPBz6SOA6Ms+scYOq2ql2Z8dmfsuoQ1tg3BQ==","signature_status":"signed_v1","signed_at":"2026-07-05T11:41:39.873515Z","signed_message":"canonical_sha256_bytes"},"source_id":"2503.03460","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:98df2fab7ac25a7011c89b98290ee2ab1cb2ae6d138271eeb7dfc1c98542d034","sha256:23ba634c2667ada4c57077ed4c9751855b5f1d9837162b5394a91626bb5034f3"],"state_sha256":"a25a162ed7fb34f1edde86109a2630e3666799a1f96a6385c0044e312b1f5708"}