{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:AUOJIZBDOEHP76PESV763RVUXP","short_pith_number":"pith:AUOJIZBD","canonical_record":{"source":{"id":"2305.03495","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-05-04T15:15:22Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"959bf99b0b128ac81d9186058b4c9df323295a0a2338b928e96c0be4e7ee4e16","abstract_canon_sha256":"03bb66f3e9d1dbd00fbe5ad028e839723a51484d33c425e15afffedcd8a85d17"},"schema_version":"1.0"},"canonical_sha256":"051c946423710efff9e4957fedc6b4bbf5bebd1643b693362b19f1ddf469192b","source":{"kind":"arxiv","id":"2305.03495","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2305.03495","created_at":"2026-07-05T07:02:28Z"},{"alias_kind":"arxiv_version","alias_value":"2305.03495v2","created_at":"2026-07-05T07:02:28Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.03495","created_at":"2026-07-05T07:02:28Z"},{"alias_kind":"pith_short_12","alias_value":"AUOJIZBDOEHP","created_at":"2026-07-05T07:02:28Z"},{"alias_kind":"pith_short_16","alias_value":"AUOJIZBDOEHP76PE","created_at":"2026-07-05T07:02:28Z"},{"alias_kind":"pith_short_8","alias_value":"AUOJIZBD","created_at":"2026-07-05T07:02:28Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:AUOJIZBDOEHP76PESV763RVUXP","target":"record","payload":{"canonical_record":{"source":{"id":"2305.03495","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-05-04T15:15:22Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"959bf99b0b128ac81d9186058b4c9df323295a0a2338b928e96c0be4e7ee4e16","abstract_canon_sha256":"03bb66f3e9d1dbd00fbe5ad028e839723a51484d33c425e15afffedcd8a85d17"},"schema_version":"1.0"},"canonical_sha256":"051c946423710efff9e4957fedc6b4bbf5bebd1643b693362b19f1ddf469192b","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:02:28.636917Z","signature_b64":"d0mDExlai2/6b0XeKQtmVArXS0n7TA2G/S1inhVkASCFh5/KqQSzwn8a5fuix+/SuVQNFBTn733L5M8zouf/Bw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"051c946423710efff9e4957fedc6b4bbf5bebd1643b693362b19f1ddf469192b","last_reissued_at":"2026-07-05T07:02:28.636416Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:02:28.636416Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2305.03495","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":"2026-07-05T07:02:28Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"x465fL2hemH+dmtUOvJkugq1N3CGrqHuNsLhhejlQLlY+bVwHFABXZN5dnxPmDOMphgl+pTKpdakE69zT4mHBQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T05:07:11.720042Z"},"content_sha256":"cc8f5f4f3d8333df9ef783b867e88234267997c3c564d6448bc576f3ef8a9e1d","schema_version":"1.0","event_id":"sha256:cc8f5f4f3d8333df9ef783b867e88234267997c3c564d6448bc576f3ef8a9e1d"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:AUOJIZBDOEHP76PESV763RVUXP","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Automatic Prompt Optimization with \"Gradient Descent\" and Beam Search","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Chenguang Zhu, Dan Iter, Jerry Li, Michael Zeng, Reid Pryzant, Yin Tat Lee","submitted_at":"2023-05-04T15:15:22Z","abstract_excerpt":"Large Language Models (LLMs) have shown impressive performance as general purpose agents, but their abilities remain highly dependent on prompts which are hand written with onerous trial-and-error effort. We propose a simple and nonparametric solution to this problem, Automatic Prompt Optimization (APO), which is inspired by numerical gradient descent to automatically improve prompts, assuming access to training data and an LLM API. The algorithm uses minibatches of data to form natural language \"gradients\" that criticize the current prompt. The gradients are then \"propagated\" into the prompt "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2305.03495","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/2305.03495/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":"2026-07-05T07:02:28Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"J18SO154WAJeDmnKjVxy090NEjBbJW5k55/XoZ6P/ygs/OsRUmaUGrcNale595nxW6B5RHtH579zf4HDh5gfAQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T05:07:11.720561Z"},"content_sha256":"b5055778787e8e8fcb21b7273bd65f1266e3920a936026f672df279062e4768f","schema_version":"1.0","event_id":"sha256:b5055778787e8e8fcb21b7273bd65f1266e3920a936026f672df279062e4768f"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/AUOJIZBDOEHP76PESV763RVUXP/bundle.json","state_url":"https://pith.science/pith/AUOJIZBDOEHP76PESV763RVUXP/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/AUOJIZBDOEHP76PESV763RVUXP/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":"2026-08-06T05:07:11Z","links":{"resolver":"https://pith.science/pith/AUOJIZBDOEHP76PESV763RVUXP","bundle":"https://pith.science/pith/AUOJIZBDOEHP76PESV763RVUXP/bundle.json","state":"https://pith.science/pith/AUOJIZBDOEHP76PESV763RVUXP/state.json","well_known_bundle":"https://pith.science/.well-known/pith/AUOJIZBDOEHP76PESV763RVUXP/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:AUOJIZBDOEHP76PESV763RVUXP","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":"03bb66f3e9d1dbd00fbe5ad028e839723a51484d33c425e15afffedcd8a85d17","cross_cats_sorted":["cs.AI","cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-05-04T15:15:22Z","title_canon_sha256":"959bf99b0b128ac81d9186058b4c9df323295a0a2338b928e96c0be4e7ee4e16"},"schema_version":"1.0","source":{"id":"2305.03495","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2305.03495","created_at":"2026-07-05T07:02:28Z"},{"alias_kind":"arxiv_version","alias_value":"2305.03495v2","created_at":"2026-07-05T07:02:28Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.03495","created_at":"2026-07-05T07:02:28Z"},{"alias_kind":"pith_short_12","alias_value":"AUOJIZBDOEHP","created_at":"2026-07-05T07:02:28Z"},{"alias_kind":"pith_short_16","alias_value":"AUOJIZBDOEHP76PE","created_at":"2026-07-05T07:02:28Z"},{"alias_kind":"pith_short_8","alias_value":"AUOJIZBD","created_at":"2026-07-05T07:02:28Z"}],"graph_snapshots":[{"event_id":"sha256:b5055778787e8e8fcb21b7273bd65f1266e3920a936026f672df279062e4768f","target":"graph","created_at":"2026-07-05T07:02:28Z","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/2305.03495/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large Language Models (LLMs) have shown impressive performance as general purpose agents, but their abilities remain highly dependent on prompts which are hand written with onerous trial-and-error effort. We propose a simple and nonparametric solution to this problem, Automatic Prompt Optimization (APO), which is inspired by numerical gradient descent to automatically improve prompts, assuming access to training data and an LLM API. The algorithm uses minibatches of data to form natural language \"gradients\" that criticize the current prompt. The gradients are then \"propagated\" into the prompt ","authors_text":"Chenguang Zhu, Dan Iter, Jerry Li, Michael Zeng, Reid Pryzant, Yin Tat Lee","cross_cats":["cs.AI","cs.LG"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-05-04T15:15:22Z","title":"Automatic Prompt Optimization with \"Gradient Descent\" and Beam Search"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2305.03495","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:cc8f5f4f3d8333df9ef783b867e88234267997c3c564d6448bc576f3ef8a9e1d","target":"record","created_at":"2026-07-05T07:02:28Z","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":"03bb66f3e9d1dbd00fbe5ad028e839723a51484d33c425e15afffedcd8a85d17","cross_cats_sorted":["cs.AI","cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-05-04T15:15:22Z","title_canon_sha256":"959bf99b0b128ac81d9186058b4c9df323295a0a2338b928e96c0be4e7ee4e16"},"schema_version":"1.0","source":{"id":"2305.03495","kind":"arxiv","version":2}},"canonical_sha256":"051c946423710efff9e4957fedc6b4bbf5bebd1643b693362b19f1ddf469192b","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"051c946423710efff9e4957fedc6b4bbf5bebd1643b693362b19f1ddf469192b","first_computed_at":"2026-07-05T07:02:28.636416Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T07:02:28.636416Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"d0mDExlai2/6b0XeKQtmVArXS0n7TA2G/S1inhVkASCFh5/KqQSzwn8a5fuix+/SuVQNFBTn733L5M8zouf/Bw==","signature_status":"signed_v1","signed_at":"2026-07-05T07:02:28.636917Z","signed_message":"canonical_sha256_bytes"},"source_id":"2305.03495","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:cc8f5f4f3d8333df9ef783b867e88234267997c3c564d6448bc576f3ef8a9e1d","sha256:b5055778787e8e8fcb21b7273bd65f1266e3920a936026f672df279062e4768f"],"state_sha256":"3dc13ddc74c93f71cb0c146536b24d2d4f497004bc086739da74f0cf55152f9f"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"fbQDja22zbEvvZnXaS+cHT8ZejbBcGx8BSEoNckSE4soojray1FPD98qGMCRpeyKAF5vm8/wSmRry180AiuCCw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-06T05:07:11.725730Z","bundle_sha256":"300f8921a4a1f38995d9ba8a76f3b8b069f8b4df7cf51032710ff32230d7940c"}}