{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:6Y7ILN7B3OPX2EUTPQGOMPKSRP","short_pith_number":"pith:6Y7ILN7B","schema_version":"1.0","canonical_sha256":"f63e85b7e1db9f7d12937c0ce63d528bd1aa0936c7dbdd9aba8e336513e45291","source":{"kind":"arxiv","id":"2106.05469","version":1},"attestation_state":"computed","paper":{"title":"Variational Information Bottleneck for Effective Low-Resource Fine-Tuning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"James Henderson, Rabeeh Karimi Mahabadi, Yonatan Belinkov","submitted_at":"2021-06-10T03:08:13Z","abstract_excerpt":"While large-scale pretrained language models have obtained impressive results when fine-tuned on a wide variety of tasks, they still often suffer from overfitting in low-resource scenarios. Since such models are general-purpose feature extractors, many of these features are inevitably irrelevant for a given target task. We propose to use Variational Information Bottleneck (VIB) to suppress irrelevant features when fine-tuning on low-resource target tasks, and show that our method successfully reduces overfitting. Moreover, we show that our VIB model finds sentence representations that are more"},"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":"2106.05469","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2021-06-10T03:08:13Z","cross_cats_sorted":[],"title_canon_sha256":"af31077a1be865cfd015447a0975845607c338414793755196f2e93dd081c7aa","abstract_canon_sha256":"7a3d0e2bc5e2339efa4bea13233eacd99d7acf3b9799b29d31be8427ea152a5a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:48:02.107055Z","signature_b64":"9P82v6JJBNq5YwmdHJfuoXTqOePe0MvhnMJQJNilhU6g2toPcPRqRPIn++WcawFtfGJT5qQQHlN963Aa6Fi/DA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f63e85b7e1db9f7d12937c0ce63d528bd1aa0936c7dbdd9aba8e336513e45291","last_reissued_at":"2026-07-05T02:48:02.106660Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:48:02.106660Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Variational Information Bottleneck for Effective Low-Resource Fine-Tuning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"James Henderson, Rabeeh Karimi Mahabadi, Yonatan Belinkov","submitted_at":"2021-06-10T03:08:13Z","abstract_excerpt":"While large-scale pretrained language models have obtained impressive results when fine-tuned on a wide variety of tasks, they still often suffer from overfitting in low-resource scenarios. Since such models are general-purpose feature extractors, many of these features are inevitably irrelevant for a given target task. We propose to use Variational Information Bottleneck (VIB) to suppress irrelevant features when fine-tuning on low-resource target tasks, and show that our method successfully reduces overfitting. Moreover, we show that our VIB model finds sentence representations that are more"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2106.05469","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/2106.05469/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":"2106.05469","created_at":"2026-07-05T02:48:02.106717+00:00"},{"alias_kind":"arxiv_version","alias_value":"2106.05469v1","created_at":"2026-07-05T02:48:02.106717+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2106.05469","created_at":"2026-07-05T02:48:02.106717+00:00"},{"alias_kind":"pith_short_12","alias_value":"6Y7ILN7B3OPX","created_at":"2026-07-05T02:48:02.106717+00:00"},{"alias_kind":"pith_short_16","alias_value":"6Y7ILN7B3OPX2EUT","created_at":"2026-07-05T02:48:02.106717+00:00"},{"alias_kind":"pith_short_8","alias_value":"6Y7ILN7B","created_at":"2026-07-05T02:48:02.106717+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.27875","citing_title":"Frequency-Aware Semantic Fusion with Gated Injection for AI-generated Image Detection","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08526","citing_title":"Skill-CMIB: Multimodal Agent Skill for Consistent Action via Conditional Multimodal Information Bottleneck","ref_index":24,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6Y7ILN7B3OPX2EUTPQGOMPKSRP","json":"https://pith.science/pith/6Y7ILN7B3OPX2EUTPQGOMPKSRP.json","graph_json":"https://pith.science/api/pith-number/6Y7ILN7B3OPX2EUTPQGOMPKSRP/graph.json","events_json":"https://pith.science/api/pith-number/6Y7ILN7B3OPX2EUTPQGOMPKSRP/events.json","paper":"https://pith.science/paper/6Y7ILN7B"},"agent_actions":{"view_html":"https://pith.science/pith/6Y7ILN7B3OPX2EUTPQGOMPKSRP","download_json":"https://pith.science/pith/6Y7ILN7B3OPX2EUTPQGOMPKSRP.json","view_paper":"https://pith.science/paper/6Y7ILN7B","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2106.05469&json=true","fetch_graph":"https://pith.science/api/pith-number/6Y7ILN7B3OPX2EUTPQGOMPKSRP/graph.json","fetch_events":"https://pith.science/api/pith-number/6Y7ILN7B3OPX2EUTPQGOMPKSRP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6Y7ILN7B3OPX2EUTPQGOMPKSRP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6Y7ILN7B3OPX2EUTPQGOMPKSRP/action/storage_attestation","attest_author":"https://pith.science/pith/6Y7ILN7B3OPX2EUTPQGOMPKSRP/action/author_attestation","sign_citation":"https://pith.science/pith/6Y7ILN7B3OPX2EUTPQGOMPKSRP/action/citation_signature","submit_replication":"https://pith.science/pith/6Y7ILN7B3OPX2EUTPQGOMPKSRP/action/replication_record"}},"created_at":"2026-07-05T02:48:02.106717+00:00","updated_at":"2026-07-05T02:48:02.106717+00:00"}