{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2019:542YWXACH7F6UJBTE74HXHEDG3","short_pith_number":"pith:542YWXAC","canonical_record":{"source":{"id":"1908.05620","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2019-08-15T16:11:45Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"5ad4d22ba05c741c18696b50f33ac97fcaa5aecffce749c4c9d9238b1f2a8266","abstract_canon_sha256":"23670d85a45606e9c4691dc61698c0401422b16781cfc0ed485e26e6d7999186"},"schema_version":"1.0"},"canonical_sha256":"ef358b5c023fcbea243327f87b9c8336e09b36a828dedb27eaba6f8c6b1e1dea","source":{"kind":"arxiv","id":"1908.05620","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1908.05620","created_at":"2026-07-04T23:56:54Z"},{"alias_kind":"arxiv_version","alias_value":"1908.05620v1","created_at":"2026-07-04T23:56:54Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1908.05620","created_at":"2026-07-04T23:56:54Z"},{"alias_kind":"pith_short_12","alias_value":"542YWXACH7F6","created_at":"2026-07-04T23:56:54Z"},{"alias_kind":"pith_short_16","alias_value":"542YWXACH7F6UJBT","created_at":"2026-07-04T23:56:54Z"},{"alias_kind":"pith_short_8","alias_value":"542YWXAC","created_at":"2026-07-04T23:56:54Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2019:542YWXACH7F6UJBTE74HXHEDG3","target":"record","payload":{"canonical_record":{"source":{"id":"1908.05620","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2019-08-15T16:11:45Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"5ad4d22ba05c741c18696b50f33ac97fcaa5aecffce749c4c9d9238b1f2a8266","abstract_canon_sha256":"23670d85a45606e9c4691dc61698c0401422b16781cfc0ed485e26e6d7999186"},"schema_version":"1.0"},"canonical_sha256":"ef358b5c023fcbea243327f87b9c8336e09b36a828dedb27eaba6f8c6b1e1dea","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-04T23:56:54.377829Z","signature_b64":"lswrdIKzqgM5r/jIngjhjdS8D7eCoYd7lIO30+0iqxeZGjjyZV/XAF5ncy5FC5pZ/rXxlaTKwjTcx7UpjBWCCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ef358b5c023fcbea243327f87b9c8336e09b36a828dedb27eaba6f8c6b1e1dea","last_reissued_at":"2026-07-04T23:56:54.377415Z","signature_status":"signed_v1","first_computed_at":"2026-07-04T23:56:54.377415Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"1908.05620","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":"2026-07-04T23:56:54Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"54GFcxfUJ+BG8OtHiEXgwNWEz+lYbhTCo941yey4Rv/h2k4p+JHqGT2myvbJUCQdME1Fu5fbZ6Zc96OmZRSmBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T05:50:39.398197Z"},"content_sha256":"3c3d789170f0697aef5db06df3dd711c65f2aedac02f7df8bcfcdb608c6ab42e","schema_version":"1.0","event_id":"sha256:3c3d789170f0697aef5db06df3dd711c65f2aedac02f7df8bcfcdb608c6ab42e"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2019:542YWXACH7F6UJBTE74HXHEDG3","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Visualizing and Understanding the Effectiveness of BERT","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Furu Wei, Ke Xu, Li Dong, Yaru Hao","submitted_at":"2019-08-15T16:11:45Z","abstract_excerpt":"Language model pre-training, such as BERT, has achieved remarkable results in many NLP tasks. However, it is unclear why the pre-training-then-fine-tuning paradigm can improve performance and generalization capability across different tasks. In this paper, we propose to visualize loss landscapes and optimization trajectories of fine-tuning BERT on specific datasets. First, we find that pre-training reaches a good initial point across downstream tasks, which leads to wider optima and easier optimization compared with training from scratch. We also demonstrate that the fine-tuning procedure is r"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1908.05620","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/1908.05620/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":"2026-07-04T23:56:54Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"7WD5eNt5L5aXiuqp586q9lXRkwvJzgLj2jsEHD6wAeCR3E4PXtpMDVlmjjeTBBJJMZK8wJ3lXYrFYtlnKijgAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T05:50:39.399123Z"},"content_sha256":"73e363dbc4b01f1aa03d57d3749e304c8129cb6ba98dbcfba3de032c6db60701","schema_version":"1.0","event_id":"sha256:73e363dbc4b01f1aa03d57d3749e304c8129cb6ba98dbcfba3de032c6db60701"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/542YWXACH7F6UJBTE74HXHEDG3/bundle.json","state_url":"https://pith.science/pith/542YWXACH7F6UJBTE74HXHEDG3/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/542YWXACH7F6UJBTE74HXHEDG3/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":"2026-08-06T05:50:39Z","links":{"resolver":"https://pith.science/pith/542YWXACH7F6UJBTE74HXHEDG3","bundle":"https://pith.science/pith/542YWXACH7F6UJBTE74HXHEDG3/bundle.json","state":"https://pith.science/pith/542YWXACH7F6UJBTE74HXHEDG3/state.json","well_known_bundle":"https://pith.science/.well-known/pith/542YWXACH7F6UJBTE74HXHEDG3/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2019:542YWXACH7F6UJBTE74HXHEDG3","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":"23670d85a45606e9c4691dc61698c0401422b16781cfc0ed485e26e6d7999186","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2019-08-15T16:11:45Z","title_canon_sha256":"5ad4d22ba05c741c18696b50f33ac97fcaa5aecffce749c4c9d9238b1f2a8266"},"schema_version":"1.0","source":{"id":"1908.05620","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1908.05620","created_at":"2026-07-04T23:56:54Z"},{"alias_kind":"arxiv_version","alias_value":"1908.05620v1","created_at":"2026-07-04T23:56:54Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1908.05620","created_at":"2026-07-04T23:56:54Z"},{"alias_kind":"pith_short_12","alias_value":"542YWXACH7F6","created_at":"2026-07-04T23:56:54Z"},{"alias_kind":"pith_short_16","alias_value":"542YWXACH7F6UJBT","created_at":"2026-07-04T23:56:54Z"},{"alias_kind":"pith_short_8","alias_value":"542YWXAC","created_at":"2026-07-04T23:56:54Z"}],"graph_snapshots":[{"event_id":"sha256:73e363dbc4b01f1aa03d57d3749e304c8129cb6ba98dbcfba3de032c6db60701","target":"graph","created_at":"2026-07-04T23:56:54Z","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/1908.05620/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Language model pre-training, such as BERT, has achieved remarkable results in many NLP tasks. However, it is unclear why the pre-training-then-fine-tuning paradigm can improve performance and generalization capability across different tasks. In this paper, we propose to visualize loss landscapes and optimization trajectories of fine-tuning BERT on specific datasets. First, we find that pre-training reaches a good initial point across downstream tasks, which leads to wider optima and easier optimization compared with training from scratch. We also demonstrate that the fine-tuning procedure is r","authors_text":"Furu Wei, Ke Xu, Li Dong, Yaru Hao","cross_cats":["cs.LG"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2019-08-15T16:11:45Z","title":"Visualizing and Understanding the Effectiveness of BERT"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1908.05620","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":"sha256:3c3d789170f0697aef5db06df3dd711c65f2aedac02f7df8bcfcdb608c6ab42e","target":"record","created_at":"2026-07-04T23:56:54Z","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":"23670d85a45606e9c4691dc61698c0401422b16781cfc0ed485e26e6d7999186","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2019-08-15T16:11:45Z","title_canon_sha256":"5ad4d22ba05c741c18696b50f33ac97fcaa5aecffce749c4c9d9238b1f2a8266"},"schema_version":"1.0","source":{"id":"1908.05620","kind":"arxiv","version":1}},"canonical_sha256":"ef358b5c023fcbea243327f87b9c8336e09b36a828dedb27eaba6f8c6b1e1dea","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"ef358b5c023fcbea243327f87b9c8336e09b36a828dedb27eaba6f8c6b1e1dea","first_computed_at":"2026-07-04T23:56:54.377415Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-04T23:56:54.377415Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"lswrdIKzqgM5r/jIngjhjdS8D7eCoYd7lIO30+0iqxeZGjjyZV/XAF5ncy5FC5pZ/rXxlaTKwjTcx7UpjBWCCA==","signature_status":"signed_v1","signed_at":"2026-07-04T23:56:54.377829Z","signed_message":"canonical_sha256_bytes"},"source_id":"1908.05620","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:3c3d789170f0697aef5db06df3dd711c65f2aedac02f7df8bcfcdb608c6ab42e","sha256:73e363dbc4b01f1aa03d57d3749e304c8129cb6ba98dbcfba3de032c6db60701"],"state_sha256":"c136a267ba560cadffe18053f89ea436f27ab763b8f3dd6277396c71f713b6f8"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"A7tSb2LIG6MLQfgcaUFtF/EUiuDVQP8Ao69JauuBCwdzzBAYPODL17ipGntaGDikG76LnQJc83sNvNTpXbVMDw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-06T05:50:39.405302Z","bundle_sha256":"041c4daa861880e881e599e0bd74bf57b9258903978aa4fa1e9efdca3b517681"}}