{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2022:IYXJA4BPRTD5H22CL66YKEGCAB","short_pith_number":"pith:IYXJA4BP","canonical_record":{"source":{"id":"2202.12024","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-02-24T11:08:02Z","cross_cats_sorted":[],"title_canon_sha256":"3a744f6be9001626107216a1138b91a69d2c45ff75b96ba710be677cb3ac36ee","abstract_canon_sha256":"00b7017d9c9c48fba4f2d20b63f0e47d373eaeed8c4bbba978b264ebaa8e474b"},"schema_version":"1.0"},"canonical_sha256":"462e90702f8cc7d3eb425fbd8510c2007022301cfbd13c75f7573fa1ce8467b7","source":{"kind":"arxiv","id":"2202.12024","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2202.12024","created_at":"2026-07-05T04:07:43Z"},{"alias_kind":"arxiv_version","alias_value":"2202.12024v2","created_at":"2026-07-05T04:07:43Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2202.12024","created_at":"2026-07-05T04:07:43Z"},{"alias_kind":"pith_short_12","alias_value":"IYXJA4BPRTD5","created_at":"2026-07-05T04:07:43Z"},{"alias_kind":"pith_short_16","alias_value":"IYXJA4BPRTD5H22C","created_at":"2026-07-05T04:07:43Z"},{"alias_kind":"pith_short_8","alias_value":"IYXJA4BP","created_at":"2026-07-05T04:07:43Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2022:IYXJA4BPRTD5H22CL66YKEGCAB","target":"record","payload":{"canonical_record":{"source":{"id":"2202.12024","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-02-24T11:08:02Z","cross_cats_sorted":[],"title_canon_sha256":"3a744f6be9001626107216a1138b91a69d2c45ff75b96ba710be677cb3ac36ee","abstract_canon_sha256":"00b7017d9c9c48fba4f2d20b63f0e47d373eaeed8c4bbba978b264ebaa8e474b"},"schema_version":"1.0"},"canonical_sha256":"462e90702f8cc7d3eb425fbd8510c2007022301cfbd13c75f7573fa1ce8467b7","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:07:43.934339Z","signature_b64":"MZvUwKt7p80kaO+90kuToZ5eqE15qFCiCvXy/ETxk6GsRNAmQeDN98eZhVa8108I/+zm8ePVzPzpwzNt5+s0AA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"462e90702f8cc7d3eb425fbd8510c2007022301cfbd13c75f7573fa1ce8467b7","last_reissued_at":"2026-07-05T04:07:43.933863Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:07:43.933863Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2202.12024","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":"2026-07-05T04:07:43Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"21DI+2dISz9MtC7YU55+od/dGIm3s8orJY42TXif9AJjKcCe5HiwHpO/Nhb0kvZX6BSxPBN1t1zwT/6x2ZZICg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T01:17:05.789783Z"},"content_sha256":"c186283303990883987806211d85f5270695bd8ac08b31b71979901864462288","schema_version":"1.0","event_id":"sha256:c186283303990883987806211d85f5270695bd8ac08b31b71979901864462288"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2022:IYXJA4BPRTD5H22CL66YKEGCAB","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"NoisyTune: A Little Noise Can Help You Finetune Pretrained Language Models Better","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chuhan Wu, Fangzhao Wu, Tao Qi, Xing Xie, Yongfeng Huang","submitted_at":"2022-02-24T11:08:02Z","abstract_excerpt":"Effectively finetuning pretrained language models (PLMs) is critical for their success in downstream tasks. However, PLMs may have risks in overfitting the pretraining tasks and data, which usually have gap with the target downstream tasks. Such gap may be difficult for existing PLM finetuning methods to overcome and lead to suboptimal performance. In this paper, we propose a very simple yet effective method named NoisyTune to help better finetune PLMs on downstream tasks by adding some noise to the parameters of PLMs before fine-tuning. More specifically, we propose a matrix-wise perturbing m"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2202.12024","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/2202.12024/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":"2026-07-05T04:07:43Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"dRi1Dk7QYPIUi/5Phz/fIxi8iXqkbCeV4eJh1aa9DUYMJHvSi5laObIE2XzuERswVZXwhAXTmvKbRVIzUoZHAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T01:17:05.790284Z"},"content_sha256":"ad11be61cb860b3781c35dab559ab4d24d01781037b6f35e09e6c6e62b517493","schema_version":"1.0","event_id":"sha256:ad11be61cb860b3781c35dab559ab4d24d01781037b6f35e09e6c6e62b517493"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/IYXJA4BPRTD5H22CL66YKEGCAB/bundle.json","state_url":"https://pith.science/pith/IYXJA4BPRTD5H22CL66YKEGCAB/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/IYXJA4BPRTD5H22CL66YKEGCAB/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":"2026-08-08T01:17:05Z","links":{"resolver":"https://pith.science/pith/IYXJA4BPRTD5H22CL66YKEGCAB","bundle":"https://pith.science/pith/IYXJA4BPRTD5H22CL66YKEGCAB/bundle.json","state":"https://pith.science/pith/IYXJA4BPRTD5H22CL66YKEGCAB/state.json","well_known_bundle":"https://pith.science/.well-known/pith/IYXJA4BPRTD5H22CL66YKEGCAB/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2022:IYXJA4BPRTD5H22CL66YKEGCAB","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":"00b7017d9c9c48fba4f2d20b63f0e47d373eaeed8c4bbba978b264ebaa8e474b","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-02-24T11:08:02Z","title_canon_sha256":"3a744f6be9001626107216a1138b91a69d2c45ff75b96ba710be677cb3ac36ee"},"schema_version":"1.0","source":{"id":"2202.12024","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2202.12024","created_at":"2026-07-05T04:07:43Z"},{"alias_kind":"arxiv_version","alias_value":"2202.12024v2","created_at":"2026-07-05T04:07:43Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2202.12024","created_at":"2026-07-05T04:07:43Z"},{"alias_kind":"pith_short_12","alias_value":"IYXJA4BPRTD5","created_at":"2026-07-05T04:07:43Z"},{"alias_kind":"pith_short_16","alias_value":"IYXJA4BPRTD5H22C","created_at":"2026-07-05T04:07:43Z"},{"alias_kind":"pith_short_8","alias_value":"IYXJA4BP","created_at":"2026-07-05T04:07:43Z"}],"graph_snapshots":[{"event_id":"sha256:ad11be61cb860b3781c35dab559ab4d24d01781037b6f35e09e6c6e62b517493","target":"graph","created_at":"2026-07-05T04:07:43Z","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/2202.12024/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Effectively finetuning pretrained language models (PLMs) is critical for their success in downstream tasks. However, PLMs may have risks in overfitting the pretraining tasks and data, which usually have gap with the target downstream tasks. Such gap may be difficult for existing PLM finetuning methods to overcome and lead to suboptimal performance. In this paper, we propose a very simple yet effective method named NoisyTune to help better finetune PLMs on downstream tasks by adding some noise to the parameters of PLMs before fine-tuning. More specifically, we propose a matrix-wise perturbing m","authors_text":"Chuhan Wu, Fangzhao Wu, Tao Qi, Xing Xie, Yongfeng Huang","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-02-24T11:08:02Z","title":"NoisyTune: A Little Noise Can Help You Finetune Pretrained Language Models Better"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2202.12024","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:c186283303990883987806211d85f5270695bd8ac08b31b71979901864462288","target":"record","created_at":"2026-07-05T04:07:43Z","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":"00b7017d9c9c48fba4f2d20b63f0e47d373eaeed8c4bbba978b264ebaa8e474b","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-02-24T11:08:02Z","title_canon_sha256":"3a744f6be9001626107216a1138b91a69d2c45ff75b96ba710be677cb3ac36ee"},"schema_version":"1.0","source":{"id":"2202.12024","kind":"arxiv","version":2}},"canonical_sha256":"462e90702f8cc7d3eb425fbd8510c2007022301cfbd13c75f7573fa1ce8467b7","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"462e90702f8cc7d3eb425fbd8510c2007022301cfbd13c75f7573fa1ce8467b7","first_computed_at":"2026-07-05T04:07:43.933863Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T04:07:43.933863Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"MZvUwKt7p80kaO+90kuToZ5eqE15qFCiCvXy/ETxk6GsRNAmQeDN98eZhVa8108I/+zm8ePVzPzpwzNt5+s0AA==","signature_status":"signed_v1","signed_at":"2026-07-05T04:07:43.934339Z","signed_message":"canonical_sha256_bytes"},"source_id":"2202.12024","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:c186283303990883987806211d85f5270695bd8ac08b31b71979901864462288","sha256:ad11be61cb860b3781c35dab559ab4d24d01781037b6f35e09e6c6e62b517493"],"state_sha256":"20958666b2d26b853c70418e3556ecc52c81d72a46965d559e1bed2027e7216b"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"1Am4oixGZC6mxD61fgQCNWjpko+Ih/4MxGBbsVKo6wHYI3KbEtzjGmzk7ybB2sYTKaELgpE0QZ3WFsU9dj86AQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-08T01:17:05.794657Z","bundle_sha256":"457e9ba43c9b3f8e8b9f4ab8e6811ae3190e84a00235ffd086742c63688045da"}}