{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:6MQZHLJADDTI6HE2INIABTE62C","short_pith_number":"pith:6MQZHLJA","schema_version":"1.0","canonical_sha256":"f32193ad2018e68f1c9a435000cc9ed0ad52bfa60c76b3fcadcd9b0089ffa3c7","source":{"kind":"arxiv","id":"2303.12314","version":4},"attestation_state":"computed","paper":{"title":"Self-supervised Meta-Prompt Learning with Meta-Gradient Regularization for Few-shot Generalization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Hongye Song, Juncheng Li, Jun Lin, Kaihang Pan, Siliang Tang, Xiaozhong Liu","submitted_at":"2023-03-22T05:04:21Z","abstract_excerpt":"Prompt tuning is a parameter-efficient method, which learns soft prompts and conditions frozen language models to perform specific downstream tasks. Though effective, prompt tuning under few-shot settings on the one hand heavily relies on a good initialization of soft prompts. On the other hand, it can easily overfit to few-shot training samples, thereby undermining generalizability. Existing works leverage pre-training or supervised meta-learning to initialize soft prompts but they fail to data-efficiently generalize to unseen downstream tasks. To address the above problems, this paper propos"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2303.12314","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-03-22T05:04:21Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"db103413f93dbe3584a138a3fe1bf043907c2e0bb8853656c3580808b07c6c16","abstract_canon_sha256":"5f412b21101b695a6a2e9998648e8b338174d7a2ab0b86d054073399e24d8484"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:03:58.251506Z","signature_b64":"tll0KWuT+BNMw9O7w9aSrVpSsFwRsSMCPHo9/L7ee3EPEjpc8ppn50vemJSzGnXzRJNPGW/2hK7/PSfiXcRdCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f32193ad2018e68f1c9a435000cc9ed0ad52bfa60c76b3fcadcd9b0089ffa3c7","last_reissued_at":"2026-07-05T07:03:58.251009Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:03:58.251009Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Self-supervised Meta-Prompt Learning with Meta-Gradient Regularization for Few-shot Generalization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Hongye Song, Juncheng Li, Jun Lin, Kaihang Pan, Siliang Tang, Xiaozhong Liu","submitted_at":"2023-03-22T05:04:21Z","abstract_excerpt":"Prompt tuning is a parameter-efficient method, which learns soft prompts and conditions frozen language models to perform specific downstream tasks. Though effective, prompt tuning under few-shot settings on the one hand heavily relies on a good initialization of soft prompts. On the other hand, it can easily overfit to few-shot training samples, thereby undermining generalizability. Existing works leverage pre-training or supervised meta-learning to initialize soft prompts but they fail to data-efficiently generalize to unseen downstream tasks. To address the above problems, this paper propos"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2303.12314","kind":"arxiv","version":4},"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/2303.12314/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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2303.12314","created_at":"2026-07-05T07:03:58.251070+00:00"},{"alias_kind":"arxiv_version","alias_value":"2303.12314v4","created_at":"2026-07-05T07:03:58.251070+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2303.12314","created_at":"2026-07-05T07:03:58.251070+00:00"},{"alias_kind":"pith_short_12","alias_value":"6MQZHLJADDTI","created_at":"2026-07-05T07:03:58.251070+00:00"},{"alias_kind":"pith_short_16","alias_value":"6MQZHLJADDTI6HE2","created_at":"2026-07-05T07:03:58.251070+00:00"},{"alias_kind":"pith_short_8","alias_value":"6MQZHLJA","created_at":"2026-07-05T07:03:58.251070+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2509.05714","citing_title":"Towards Meta-Cognitive Knowledge Editing for Multimodal LLMs","ref_index":45,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6MQZHLJADDTI6HE2INIABTE62C","json":"https://pith.science/pith/6MQZHLJADDTI6HE2INIABTE62C.json","graph_json":"https://pith.science/api/pith-number/6MQZHLJADDTI6HE2INIABTE62C/graph.json","events_json":"https://pith.science/api/pith-number/6MQZHLJADDTI6HE2INIABTE62C/events.json","paper":"https://pith.science/paper/6MQZHLJA"},"agent_actions":{"view_html":"https://pith.science/pith/6MQZHLJADDTI6HE2INIABTE62C","download_json":"https://pith.science/pith/6MQZHLJADDTI6HE2INIABTE62C.json","view_paper":"https://pith.science/paper/6MQZHLJA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2303.12314&json=true","fetch_graph":"https://pith.science/api/pith-number/6MQZHLJADDTI6HE2INIABTE62C/graph.json","fetch_events":"https://pith.science/api/pith-number/6MQZHLJADDTI6HE2INIABTE62C/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6MQZHLJADDTI6HE2INIABTE62C/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6MQZHLJADDTI6HE2INIABTE62C/action/storage_attestation","attest_author":"https://pith.science/pith/6MQZHLJADDTI6HE2INIABTE62C/action/author_attestation","sign_citation":"https://pith.science/pith/6MQZHLJADDTI6HE2INIABTE62C/action/citation_signature","submit_replication":"https://pith.science/pith/6MQZHLJADDTI6HE2INIABTE62C/action/replication_record"}},"created_at":"2026-07-05T07:03:58.251070+00:00","updated_at":"2026-07-05T07:03:58.251070+00:00"}