{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2022:S4XOO6TZIL42HFYHMRNOULKIBS","short_pith_number":"pith:S4XOO6TZ","canonical_record":{"source":{"id":"2205.12558","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-05-25T08:09:03Z","cross_cats_sorted":[],"title_canon_sha256":"5f6e4cd21499d6ca894af328ea5bc04beb2df86c441212fcb54cf5ba94ee1ee2","abstract_canon_sha256":"2f15efb1797b9414b26ecb98ff0942c91cbf95b43e871dad36f436ab368dee19"},"schema_version":"1.0"},"canonical_sha256":"972ee77a7942f9a39707645aea2d480cbf181e1cd95e3617ef3960d5ff508b98","source":{"kind":"arxiv","id":"2205.12558","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2205.12558","created_at":"2026-07-05T05:13:56Z"},{"alias_kind":"arxiv_version","alias_value":"2205.12558v2","created_at":"2026-07-05T05:13:56Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2205.12558","created_at":"2026-07-05T05:13:56Z"},{"alias_kind":"pith_short_12","alias_value":"S4XOO6TZIL42","created_at":"2026-07-05T05:13:56Z"},{"alias_kind":"pith_short_16","alias_value":"S4XOO6TZIL42HFYH","created_at":"2026-07-05T05:13:56Z"},{"alias_kind":"pith_short_8","alias_value":"S4XOO6TZ","created_at":"2026-07-05T05:13:56Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2022:S4XOO6TZIL42HFYHMRNOULKIBS","target":"record","payload":{"canonical_record":{"source":{"id":"2205.12558","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-05-25T08:09:03Z","cross_cats_sorted":[],"title_canon_sha256":"5f6e4cd21499d6ca894af328ea5bc04beb2df86c441212fcb54cf5ba94ee1ee2","abstract_canon_sha256":"2f15efb1797b9414b26ecb98ff0942c91cbf95b43e871dad36f436ab368dee19"},"schema_version":"1.0"},"canonical_sha256":"972ee77a7942f9a39707645aea2d480cbf181e1cd95e3617ef3960d5ff508b98","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:13:56.014384Z","signature_b64":"COz3G2tTJH0qzao4Mjbg2h8jBiwQc7mPc/HQqlsqj8Fcewav1i9jhbMZ6LFG18o0us7j7u7RkE6GO4pP30I0CA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"972ee77a7942f9a39707645aea2d480cbf181e1cd95e3617ef3960d5ff508b98","last_reissued_at":"2026-07-05T05:13:56.013877Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:13:56.013877Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2205.12558","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T05:13:56Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Vqqv3SZZ7WRINQfl3KqV6rViDOezznGbxxxv5yIk1VYdfk3Oh+JFs79q7GRJ6YsV/MBnjfJ5CCWuCJyheeMiBA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T00:23:57.424603Z"},"content_sha256":"6280d07107b8351b7299ae0c24e6485e8c295b196bdc6de204b7a1114c2edce6","schema_version":"1.0","event_id":"sha256:6280d07107b8351b7299ae0c24e6485e8c295b196bdc6de204b7a1114c2edce6"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2022:S4XOO6TZIL42HFYHMRNOULKIBS","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Gradient-Based Constrained Sampling from Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Biswajit Paria, Sachin Kumar, Yulia Tsvetkov","submitted_at":"2022-05-25T08:09:03Z","abstract_excerpt":"Large pretrained language models generate fluent text but are notoriously hard to controllably sample from. In this work, we study constrained sampling from such language models: generating text that satisfies user-defined constraints, while maintaining fluency and the model's performance in a downstream task. We propose MuCoLa -- a sampling procedure that combines the log-likelihood of the language model with arbitrary (differentiable) constraints in a single energy function, and then generates samples in a non-autoregressive manner. Specifically, it initializes the entire output sequence wit"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2205.12558","kind":"arxiv","version":2},"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/2205.12558/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":"2026-07-05T05:13:56Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"XnAeTAePVemkMy+B+UiBJlwkUb27mO6nMQzQ5AQzgOnj5QmzUSenxcrnXTeupTqLzMpe6ppHRtWW0OTuiKpuBg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T00:23:57.425189Z"},"content_sha256":"8a5a7fab7e0648ae983f9d3e77d1090033dbfd0dded4e603e459c2a90acb6ca5","schema_version":"1.0","event_id":"sha256:8a5a7fab7e0648ae983f9d3e77d1090033dbfd0dded4e603e459c2a90acb6ca5"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/S4XOO6TZIL42HFYHMRNOULKIBS/bundle.json","state_url":"https://pith.science/pith/S4XOO6TZIL42HFYHMRNOULKIBS/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/S4XOO6TZIL42HFYHMRNOULKIBS/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":"2026-08-09T00:23:57Z","links":{"resolver":"https://pith.science/pith/S4XOO6TZIL42HFYHMRNOULKIBS","bundle":"https://pith.science/pith/S4XOO6TZIL42HFYHMRNOULKIBS/bundle.json","state":"https://pith.science/pith/S4XOO6TZIL42HFYHMRNOULKIBS/state.json","well_known_bundle":"https://pith.science/.well-known/pith/S4XOO6TZIL42HFYHMRNOULKIBS/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2022:S4XOO6TZIL42HFYHMRNOULKIBS","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":"2f15efb1797b9414b26ecb98ff0942c91cbf95b43e871dad36f436ab368dee19","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-05-25T08:09:03Z","title_canon_sha256":"5f6e4cd21499d6ca894af328ea5bc04beb2df86c441212fcb54cf5ba94ee1ee2"},"schema_version":"1.0","source":{"id":"2205.12558","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2205.12558","created_at":"2026-07-05T05:13:56Z"},{"alias_kind":"arxiv_version","alias_value":"2205.12558v2","created_at":"2026-07-05T05:13:56Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2205.12558","created_at":"2026-07-05T05:13:56Z"},{"alias_kind":"pith_short_12","alias_value":"S4XOO6TZIL42","created_at":"2026-07-05T05:13:56Z"},{"alias_kind":"pith_short_16","alias_value":"S4XOO6TZIL42HFYH","created_at":"2026-07-05T05:13:56Z"},{"alias_kind":"pith_short_8","alias_value":"S4XOO6TZ","created_at":"2026-07-05T05:13:56Z"}],"graph_snapshots":[{"event_id":"sha256:8a5a7fab7e0648ae983f9d3e77d1090033dbfd0dded4e603e459c2a90acb6ca5","target":"graph","created_at":"2026-07-05T05:13:56Z","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/2205.12558/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large pretrained language models generate fluent text but are notoriously hard to controllably sample from. In this work, we study constrained sampling from such language models: generating text that satisfies user-defined constraints, while maintaining fluency and the model's performance in a downstream task. We propose MuCoLa -- a sampling procedure that combines the log-likelihood of the language model with arbitrary (differentiable) constraints in a single energy function, and then generates samples in a non-autoregressive manner. Specifically, it initializes the entire output sequence wit","authors_text":"Biswajit Paria, Sachin Kumar, Yulia Tsvetkov","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-05-25T08:09:03Z","title":"Gradient-Based Constrained Sampling from Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2205.12558","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":"sha256:6280d07107b8351b7299ae0c24e6485e8c295b196bdc6de204b7a1114c2edce6","target":"record","created_at":"2026-07-05T05:13:56Z","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":"2f15efb1797b9414b26ecb98ff0942c91cbf95b43e871dad36f436ab368dee19","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-05-25T08:09:03Z","title_canon_sha256":"5f6e4cd21499d6ca894af328ea5bc04beb2df86c441212fcb54cf5ba94ee1ee2"},"schema_version":"1.0","source":{"id":"2205.12558","kind":"arxiv","version":2}},"canonical_sha256":"972ee77a7942f9a39707645aea2d480cbf181e1cd95e3617ef3960d5ff508b98","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"972ee77a7942f9a39707645aea2d480cbf181e1cd95e3617ef3960d5ff508b98","first_computed_at":"2026-07-05T05:13:56.013877Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T05:13:56.013877Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"COz3G2tTJH0qzao4Mjbg2h8jBiwQc7mPc/HQqlsqj8Fcewav1i9jhbMZ6LFG18o0us7j7u7RkE6GO4pP30I0CA==","signature_status":"signed_v1","signed_at":"2026-07-05T05:13:56.014384Z","signed_message":"canonical_sha256_bytes"},"source_id":"2205.12558","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:6280d07107b8351b7299ae0c24e6485e8c295b196bdc6de204b7a1114c2edce6","sha256:8a5a7fab7e0648ae983f9d3e77d1090033dbfd0dded4e603e459c2a90acb6ca5"],"state_sha256":"71fb271e3721c277af64f5f9d914845a38c7f78689730ca83a084778bf594aa0"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"JHGCyesYxA+TCs9XYug1BDBd7Wkce25S3PRvIyVAz63iPG0NJlE2LY6ZHzEhwBym/mUCxtNkHqCTJUfMXku0Dw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-09T00:23:57.428519Z","bundle_sha256":"3d352d722b99d024479d2a6614c023d178078954b59ce16c1f50e352e82f27e7"}}