{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:JQ7B4CGAT4CTVRZ4423YGHUAYJ","short_pith_number":"pith:JQ7B4CGA","canonical_record":{"source":{"id":"2605.23244","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-22T05:25:00Z","cross_cats_sorted":[],"title_canon_sha256":"01ca826916e334530cc27f3724ad1a0438f8cc4dc4e574f5289c34a08258f7f6","abstract_canon_sha256":"9cba82e4eb54a9643f7128811a523b2704d282af607f9fcbbb58970b223e32da"},"schema_version":"1.0"},"canonical_sha256":"4c3e1e08c09f053ac73ce6b7831e80c25c700ccbe2545dd8526e8aa8f8ef04e4","source":{"kind":"arxiv","id":"2605.23244","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.23244","created_at":"2026-05-25T02:01:45Z"},{"alias_kind":"arxiv_version","alias_value":"2605.23244v1","created_at":"2026-05-25T02:01:45Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.23244","created_at":"2026-05-25T02:01:45Z"},{"alias_kind":"pith_short_12","alias_value":"JQ7B4CGAT4CT","created_at":"2026-05-25T02:01:45Z"},{"alias_kind":"pith_short_16","alias_value":"JQ7B4CGAT4CTVRZ4","created_at":"2026-05-25T02:01:45Z"},{"alias_kind":"pith_short_8","alias_value":"JQ7B4CGA","created_at":"2026-05-25T02:01:45Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:JQ7B4CGAT4CTVRZ4423YGHUAYJ","target":"record","payload":{"canonical_record":{"source":{"id":"2605.23244","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-22T05:25:00Z","cross_cats_sorted":[],"title_canon_sha256":"01ca826916e334530cc27f3724ad1a0438f8cc4dc4e574f5289c34a08258f7f6","abstract_canon_sha256":"9cba82e4eb54a9643f7128811a523b2704d282af607f9fcbbb58970b223e32da"},"schema_version":"1.0"},"canonical_sha256":"4c3e1e08c09f053ac73ce6b7831e80c25c700ccbe2545dd8526e8aa8f8ef04e4","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-25T02:01:45.407488Z","signature_b64":"mFEreHpQmq29lhmvJ8aMAP8q0c0nOoPrm6GQ83wb0IA4+TCk51f3cGuvEs4buW85FdyInMaTl0KMAh7fyNzRBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4c3e1e08c09f053ac73ce6b7831e80c25c700ccbe2545dd8526e8aa8f8ef04e4","last_reissued_at":"2026-05-25T02:01:45.406746Z","signature_status":"signed_v1","first_computed_at":"2026-05-25T02:01:45.406746Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.23244","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":"2026-05-25T02:01:45Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"QVUSncqAjj03xUaK/tS3JsAnzIUaOHAmPNjcJBVJzW1fCbO3pgFqmzq5JzaRfC/xpJRi3TzTjdy1oAdMF9alAA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-25T22:55:31.595647Z"},"content_sha256":"6c194717923fddba3c51bdf5d8ec44af979b26429e12b367a9519783f29c530c","schema_version":"1.0","event_id":"sha256:6c194717923fddba3c51bdf5d8ec44af979b26429e12b367a9519783f29c530c"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:JQ7B4CGAT4CTVRZ4423YGHUAYJ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Convex Optimization for Alignment and Preference Learning on a Single GPU","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Mert Pilanci, Miria Feng","submitted_at":"2026-05-22T05:25:00Z","abstract_excerpt":"Fine-tuning large language models (LLMs) to align with human preferences has driven the success of systems such as Gemini and ChatGPT. However, approaches like Reinforcement Learning from Human Feedback (RLHF) remain computationally expensive and complex. Direct Preference Optimization (DPO) offers a simpler alternative but has limitations such as inconsistent ranking accuracy, high dependence on GPU resources, and expensive hyperparameter tuning. We propose the Convex Optimization for Alignment and Preference Learning Algorithm (COALA): a novel lightweight strategy with strong theoretical gua"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.23244","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/2605.23244/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":"2026-05-25T02:01:45Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"PfNomRz0qKP7I3ySPlu49KOod5PwQqUGe23FkjNgsl1a31c4lOqik2Reqj3cfZFWbOe4hdqG8f9Ebcj9OT1GDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-25T22:55:31.596265Z"},"content_sha256":"a6d99b9bebf6955d9fb40559496d0bc41d2fd4129c377a2963ddd5f38e9725b1","schema_version":"1.0","event_id":"sha256:a6d99b9bebf6955d9fb40559496d0bc41d2fd4129c377a2963ddd5f38e9725b1"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/JQ7B4CGAT4CTVRZ4423YGHUAYJ/bundle.json","state_url":"https://pith.science/pith/JQ7B4CGAT4CTVRZ4423YGHUAYJ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/JQ7B4CGAT4CTVRZ4423YGHUAYJ/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":"2026-05-25T22:55:31Z","links":{"resolver":"https://pith.science/pith/JQ7B4CGAT4CTVRZ4423YGHUAYJ","bundle":"https://pith.science/pith/JQ7B4CGAT4CTVRZ4423YGHUAYJ/bundle.json","state":"https://pith.science/pith/JQ7B4CGAT4CTVRZ4423YGHUAYJ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/JQ7B4CGAT4CTVRZ4423YGHUAYJ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:JQ7B4CGAT4CTVRZ4423YGHUAYJ","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":"9cba82e4eb54a9643f7128811a523b2704d282af607f9fcbbb58970b223e32da","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-22T05:25:00Z","title_canon_sha256":"01ca826916e334530cc27f3724ad1a0438f8cc4dc4e574f5289c34a08258f7f6"},"schema_version":"1.0","source":{"id":"2605.23244","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.23244","created_at":"2026-05-25T02:01:45Z"},{"alias_kind":"arxiv_version","alias_value":"2605.23244v1","created_at":"2026-05-25T02:01:45Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.23244","created_at":"2026-05-25T02:01:45Z"},{"alias_kind":"pith_short_12","alias_value":"JQ7B4CGAT4CT","created_at":"2026-05-25T02:01:45Z"},{"alias_kind":"pith_short_16","alias_value":"JQ7B4CGAT4CTVRZ4","created_at":"2026-05-25T02:01:45Z"},{"alias_kind":"pith_short_8","alias_value":"JQ7B4CGA","created_at":"2026-05-25T02:01:45Z"}],"graph_snapshots":[{"event_id":"sha256:a6d99b9bebf6955d9fb40559496d0bc41d2fd4129c377a2963ddd5f38e9725b1","target":"graph","created_at":"2026-05-25T02:01:45Z","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/2605.23244/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Fine-tuning large language models (LLMs) to align with human preferences has driven the success of systems such as Gemini and ChatGPT. However, approaches like Reinforcement Learning from Human Feedback (RLHF) remain computationally expensive and complex. Direct Preference Optimization (DPO) offers a simpler alternative but has limitations such as inconsistent ranking accuracy, high dependence on GPU resources, and expensive hyperparameter tuning. We propose the Convex Optimization for Alignment and Preference Learning Algorithm (COALA): a novel lightweight strategy with strong theoretical gua","authors_text":"Mert Pilanci, Miria Feng","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-22T05:25:00Z","title":"Convex Optimization for Alignment and Preference Learning on a Single GPU"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.23244","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":"sha256:6c194717923fddba3c51bdf5d8ec44af979b26429e12b367a9519783f29c530c","target":"record","created_at":"2026-05-25T02:01:45Z","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":"9cba82e4eb54a9643f7128811a523b2704d282af607f9fcbbb58970b223e32da","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-22T05:25:00Z","title_canon_sha256":"01ca826916e334530cc27f3724ad1a0438f8cc4dc4e574f5289c34a08258f7f6"},"schema_version":"1.0","source":{"id":"2605.23244","kind":"arxiv","version":1}},"canonical_sha256":"4c3e1e08c09f053ac73ce6b7831e80c25c700ccbe2545dd8526e8aa8f8ef04e4","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"4c3e1e08c09f053ac73ce6b7831e80c25c700ccbe2545dd8526e8aa8f8ef04e4","first_computed_at":"2026-05-25T02:01:45.406746Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-25T02:01:45.406746Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"mFEreHpQmq29lhmvJ8aMAP8q0c0nOoPrm6GQ83wb0IA4+TCk51f3cGuvEs4buW85FdyInMaTl0KMAh7fyNzRBg==","signature_status":"signed_v1","signed_at":"2026-05-25T02:01:45.407488Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.23244","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:6c194717923fddba3c51bdf5d8ec44af979b26429e12b367a9519783f29c530c","sha256:a6d99b9bebf6955d9fb40559496d0bc41d2fd4129c377a2963ddd5f38e9725b1"],"state_sha256":"450a3556f90475a4765853736a53fc1db00fc32fd33624996fe7e67251ffe8de"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Mt+ROIpKGIt6OQj4oZcrLFGm/9ZEnuuz/OWhi5AtVhoWU3DoT+MVDZ/ec5p8Hvc9nny0EhbcSF1CDyIDKNlgAQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-25T22:55:31.600054Z","bundle_sha256":"c0581cc05b2fa253b5916a06ad1593fd4c23ba3339ff7a68f357ac32b7e1a4a0"}}