{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:BTJRI6D7ONU7JDARGKQVTQNTFW","short_pith_number":"pith:BTJRI6D7","canonical_record":{"source":{"id":"2404.17546","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-04-26T17:18:32Z","cross_cats_sorted":["cs.AI","cs.CL","stat.ML"],"title_canon_sha256":"f68580284269992fcbaea7b57994a65f538d8215c68d0f2352c1043b8125ac84","abstract_canon_sha256":"c791e1f156805016a5d962dba628b6ff7e6e38a6e0583cb3cd8245014d97a435"},"schema_version":"1.0"},"canonical_sha256":"0cd314787f7369f48c1132a159c1b32d9fee363ea24d3dc9207fcb32e3565b31","source":{"kind":"arxiv","id":"2404.17546","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2404.17546","created_at":"2026-07-05T08:13:34Z"},{"alias_kind":"arxiv_version","alias_value":"2404.17546v1","created_at":"2026-07-05T08:13:34Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2404.17546","created_at":"2026-07-05T08:13:34Z"},{"alias_kind":"pith_short_12","alias_value":"BTJRI6D7ONU7","created_at":"2026-07-05T08:13:34Z"},{"alias_kind":"pith_short_16","alias_value":"BTJRI6D7ONU7JDAR","created_at":"2026-07-05T08:13:34Z"},{"alias_kind":"pith_short_8","alias_value":"BTJRI6D7","created_at":"2026-07-05T08:13:34Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:BTJRI6D7ONU7JDARGKQVTQNTFW","target":"record","payload":{"canonical_record":{"source":{"id":"2404.17546","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-04-26T17:18:32Z","cross_cats_sorted":["cs.AI","cs.CL","stat.ML"],"title_canon_sha256":"f68580284269992fcbaea7b57994a65f538d8215c68d0f2352c1043b8125ac84","abstract_canon_sha256":"c791e1f156805016a5d962dba628b6ff7e6e38a6e0583cb3cd8245014d97a435"},"schema_version":"1.0"},"canonical_sha256":"0cd314787f7369f48c1132a159c1b32d9fee363ea24d3dc9207fcb32e3565b31","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:13:34.115046Z","signature_b64":"nkc5SAfXcCIYmkl4JVkTBY07PCT3RgRQxnzZw/GYji3rpHfHDRkbcmqT22zeFwOMozkavftVGGIucvKAPc+/BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0cd314787f7369f48c1132a159c1b32d9fee363ea24d3dc9207fcb32e3565b31","last_reissued_at":"2026-07-05T08:13:34.114567Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:13:34.114567Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2404.17546","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":"2026-07-05T08:13:34Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"l7gngE97q2Q5oUtGw0YwTuPBe6SIufTvI6jBojJxYSrn1wV1WybTbRRruKQhRnHabxeeOfhyxkFva2TdV9yzAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-05T12:52:21.022617Z"},"content_sha256":"38b91ebea774c3baf732b43a0cadaeb9c769b27f2f043b022a74eeb8ae73b900","schema_version":"1.0","event_id":"sha256:38b91ebea774c3baf732b43a0cadaeb9c769b27f2f043b022a74eeb8ae73b900"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:BTJRI6D7ONU7JDARGKQVTQNTFW","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Probabilistic Inference in Language Models via Twisted Sequential Monte Carlo","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Alireza Makhzani, Rob Brekelmans, Roger Grosse, Stephen Zhao","submitted_at":"2024-04-26T17:18:32Z","abstract_excerpt":"Numerous capability and safety techniques of Large Language Models (LLMs), including RLHF, automated red-teaming, prompt engineering, and infilling, can be cast as sampling from an unnormalized target distribution defined by a given reward or potential function over the full sequence. In this work, we leverage the rich toolkit of Sequential Monte Carlo (SMC) for these probabilistic inference problems. In particular, we use learned twist functions to estimate the expected future value of the potential at each timestep, which enables us to focus inference-time computation on promising partial se"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2404.17546","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/2404.17546/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":"2026-07-05T08:13:34Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"t1o54X1eqZXu2oJ1jC+9mVJB+t4I85wfVXwYj3LKjXBjKAMjJqQbCX+I0oCgUoWZ7GMWJERvyUwl8UWOre6nDw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-05T12:52:21.022931Z"},"content_sha256":"279415d30836f1e8470e441fbc638a99f1a887370ee454d556117c6ab833d00c","schema_version":"1.0","event_id":"sha256:279415d30836f1e8470e441fbc638a99f1a887370ee454d556117c6ab833d00c"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/BTJRI6D7ONU7JDARGKQVTQNTFW/bundle.json","state_url":"https://pith.science/pith/BTJRI6D7ONU7JDARGKQVTQNTFW/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/BTJRI6D7ONU7JDARGKQVTQNTFW/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":"2026-08-05T12:52:21Z","links":{"resolver":"https://pith.science/pith/BTJRI6D7ONU7JDARGKQVTQNTFW","bundle":"https://pith.science/pith/BTJRI6D7ONU7JDARGKQVTQNTFW/bundle.json","state":"https://pith.science/pith/BTJRI6D7ONU7JDARGKQVTQNTFW/state.json","well_known_bundle":"https://pith.science/.well-known/pith/BTJRI6D7ONU7JDARGKQVTQNTFW/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:BTJRI6D7ONU7JDARGKQVTQNTFW","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":"c791e1f156805016a5d962dba628b6ff7e6e38a6e0583cb3cd8245014d97a435","cross_cats_sorted":["cs.AI","cs.CL","stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-04-26T17:18:32Z","title_canon_sha256":"f68580284269992fcbaea7b57994a65f538d8215c68d0f2352c1043b8125ac84"},"schema_version":"1.0","source":{"id":"2404.17546","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2404.17546","created_at":"2026-07-05T08:13:34Z"},{"alias_kind":"arxiv_version","alias_value":"2404.17546v1","created_at":"2026-07-05T08:13:34Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2404.17546","created_at":"2026-07-05T08:13:34Z"},{"alias_kind":"pith_short_12","alias_value":"BTJRI6D7ONU7","created_at":"2026-07-05T08:13:34Z"},{"alias_kind":"pith_short_16","alias_value":"BTJRI6D7ONU7JDAR","created_at":"2026-07-05T08:13:34Z"},{"alias_kind":"pith_short_8","alias_value":"BTJRI6D7","created_at":"2026-07-05T08:13:34Z"}],"graph_snapshots":[{"event_id":"sha256:279415d30836f1e8470e441fbc638a99f1a887370ee454d556117c6ab833d00c","target":"graph","created_at":"2026-07-05T08:13:34Z","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/2404.17546/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Numerous capability and safety techniques of Large Language Models (LLMs), including RLHF, automated red-teaming, prompt engineering, and infilling, can be cast as sampling from an unnormalized target distribution defined by a given reward or potential function over the full sequence. In this work, we leverage the rich toolkit of Sequential Monte Carlo (SMC) for these probabilistic inference problems. In particular, we use learned twist functions to estimate the expected future value of the potential at each timestep, which enables us to focus inference-time computation on promising partial se","authors_text":"Alireza Makhzani, Rob Brekelmans, Roger Grosse, Stephen Zhao","cross_cats":["cs.AI","cs.CL","stat.ML"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-04-26T17:18:32Z","title":"Probabilistic Inference in Language Models via Twisted Sequential Monte Carlo"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2404.17546","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":"sha256:38b91ebea774c3baf732b43a0cadaeb9c769b27f2f043b022a74eeb8ae73b900","target":"record","created_at":"2026-07-05T08:13:34Z","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":"c791e1f156805016a5d962dba628b6ff7e6e38a6e0583cb3cd8245014d97a435","cross_cats_sorted":["cs.AI","cs.CL","stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-04-26T17:18:32Z","title_canon_sha256":"f68580284269992fcbaea7b57994a65f538d8215c68d0f2352c1043b8125ac84"},"schema_version":"1.0","source":{"id":"2404.17546","kind":"arxiv","version":1}},"canonical_sha256":"0cd314787f7369f48c1132a159c1b32d9fee363ea24d3dc9207fcb32e3565b31","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"0cd314787f7369f48c1132a159c1b32d9fee363ea24d3dc9207fcb32e3565b31","first_computed_at":"2026-07-05T08:13:34.114567Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:13:34.114567Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"nkc5SAfXcCIYmkl4JVkTBY07PCT3RgRQxnzZw/GYji3rpHfHDRkbcmqT22zeFwOMozkavftVGGIucvKAPc+/BQ==","signature_status":"signed_v1","signed_at":"2026-07-05T08:13:34.115046Z","signed_message":"canonical_sha256_bytes"},"source_id":"2404.17546","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:38b91ebea774c3baf732b43a0cadaeb9c769b27f2f043b022a74eeb8ae73b900","sha256:279415d30836f1e8470e441fbc638a99f1a887370ee454d556117c6ab833d00c"],"state_sha256":"eec4190169166ff9e2720e3f721559bccf55fe891ed60fc5ba50d802978de1fd"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"rBEn3GJdlNfTqW7FBwB053VcWSA9FVt1FogsU7FkYhIFUSBtIgklSAHn1pda4Xap38tgrlrNqLT/Bx7HXPTmAQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-05T12:52:21.026330Z","bundle_sha256":"3d3fee028025f563fa722c6bbb7827e0f74a7bd71e6805704bd2a3b5ae2556b6"}}