{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2019:774XCGOOEOOUBHY6FDK457BI2T","short_pith_number":"pith:774XCGOO","canonical_record":{"source":{"id":"1908.05672","kind":"arxiv","version":5},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2019-08-15T03:33:50Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"14b3dc32f816ebccbf7f796d040affbbe568bf642a86d7b707e112238ea6761f","abstract_canon_sha256":"f3e7b313c4d2ef3ee12dcb5a0d0aa307649419e598262dc37d54724bcd8dea9d"},"schema_version":"1.0"},"canonical_sha256":"fff97119ce239d409f1e28d5cefc28d4e139eb532c8cc42dc555e2216950f422","source":{"kind":"arxiv","id":"1908.05672","version":5},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1908.05672","created_at":"2026-07-05T04:32:46Z"},{"alias_kind":"arxiv_version","alias_value":"1908.05672v5","created_at":"2026-07-05T04:32:46Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1908.05672","created_at":"2026-07-05T04:32:46Z"},{"alias_kind":"pith_short_12","alias_value":"774XCGOOEOOU","created_at":"2026-07-05T04:32:46Z"},{"alias_kind":"pith_short_16","alias_value":"774XCGOOEOOUBHY6","created_at":"2026-07-05T04:32:46Z"},{"alias_kind":"pith_short_8","alias_value":"774XCGOO","created_at":"2026-07-05T04:32:46Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2019:774XCGOOEOOUBHY6FDK457BI2T","target":"record","payload":{"canonical_record":{"source":{"id":"1908.05672","kind":"arxiv","version":5},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2019-08-15T03:33:50Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"14b3dc32f816ebccbf7f796d040affbbe568bf642a86d7b707e112238ea6761f","abstract_canon_sha256":"f3e7b313c4d2ef3ee12dcb5a0d0aa307649419e598262dc37d54724bcd8dea9d"},"schema_version":"1.0"},"canonical_sha256":"fff97119ce239d409f1e28d5cefc28d4e139eb532c8cc42dc555e2216950f422","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:32:46.323435Z","signature_b64":"OaiwDyDkdLIdiJqMG+kfMpZDJH9pPmCdEm46ok8swRtJO5T7x5TyEwCNWsAmcU/c8OE9BMW/sAxUWDIyZhFZBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fff97119ce239d409f1e28d5cefc28d4e139eb532c8cc42dc555e2216950f422","last_reissued_at":"2026-07-05T04:32:46.322934Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:32:46.322934Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"1908.05672","source_version":5,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T04:32:46Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"psG8FytaSOV7PjkK1hlRnFBdochsVhuLQZ5ByIF/zoUKwaFfa9s9VEBftgq1QgTjNKSIOmz4MhKXvPE1DmxpBQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-07T12:11:55.277164Z"},"content_sha256":"b25ffa684c9ff834fb2d2273d83eb87b7f7ca3f157c901808fbced65278d385d","schema_version":"1.0","event_id":"sha256:b25ffa684c9ff834fb2d2273d83eb87b7f7ca3f157c901808fbced65278d385d"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2019:774XCGOOEOOUBHY6FDK457BI2T","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Towards Making the Most of BERT in Neural Machine Translation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Chengqi Zhao, Hao Zhou, Jiacheng Yang, Lei Li, Mingxuan Wang, Weinan Zhang, Yong Yu","submitted_at":"2019-08-15T03:33:50Z","abstract_excerpt":"GPT-2 and BERT demonstrate the effectiveness of using pre-trained language models (LMs) on various natural language processing tasks. However, LM fine-tuning often suffers from catastrophic forgetting when applied to resource-rich tasks. In this work, we introduce a concerted training framework (CTNMT) that is the key to integrate the pre-trained LMs to neural machine translation (NMT). Our proposed CTNMT consists of three techniques: a) asymptotic distillation to ensure that the NMT model can retain the previous pre-trained knowledge; b) a dynamic switching gate to avoid catastrophic forgetti"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1908.05672","kind":"arxiv","version":5},"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/1908.05672/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":"2026-07-05T04:32:46Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"6AeT6vtd9gqnGpQqYCd5mlWv49deMQ6nw8h7RyNawNQMTZhGqgZBKEHq9eeEvmTwzb78iT4/JpuINgU8lHHQCg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-07T12:11:55.277804Z"},"content_sha256":"92f88ad690f97f787ffb454a428a598056ab2c17a2e1a0278786864cb5467d05","schema_version":"1.0","event_id":"sha256:92f88ad690f97f787ffb454a428a598056ab2c17a2e1a0278786864cb5467d05"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/774XCGOOEOOUBHY6FDK457BI2T/bundle.json","state_url":"https://pith.science/pith/774XCGOOEOOUBHY6FDK457BI2T/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/774XCGOOEOOUBHY6FDK457BI2T/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":"2026-07-07T12:11:55Z","links":{"resolver":"https://pith.science/pith/774XCGOOEOOUBHY6FDK457BI2T","bundle":"https://pith.science/pith/774XCGOOEOOUBHY6FDK457BI2T/bundle.json","state":"https://pith.science/pith/774XCGOOEOOUBHY6FDK457BI2T/state.json","well_known_bundle":"https://pith.science/.well-known/pith/774XCGOOEOOUBHY6FDK457BI2T/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2019:774XCGOOEOOUBHY6FDK457BI2T","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":"f3e7b313c4d2ef3ee12dcb5a0d0aa307649419e598262dc37d54724bcd8dea9d","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2019-08-15T03:33:50Z","title_canon_sha256":"14b3dc32f816ebccbf7f796d040affbbe568bf642a86d7b707e112238ea6761f"},"schema_version":"1.0","source":{"id":"1908.05672","kind":"arxiv","version":5}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1908.05672","created_at":"2026-07-05T04:32:46Z"},{"alias_kind":"arxiv_version","alias_value":"1908.05672v5","created_at":"2026-07-05T04:32:46Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1908.05672","created_at":"2026-07-05T04:32:46Z"},{"alias_kind":"pith_short_12","alias_value":"774XCGOOEOOU","created_at":"2026-07-05T04:32:46Z"},{"alias_kind":"pith_short_16","alias_value":"774XCGOOEOOUBHY6","created_at":"2026-07-05T04:32:46Z"},{"alias_kind":"pith_short_8","alias_value":"774XCGOO","created_at":"2026-07-05T04:32:46Z"}],"graph_snapshots":[{"event_id":"sha256:92f88ad690f97f787ffb454a428a598056ab2c17a2e1a0278786864cb5467d05","target":"graph","created_at":"2026-07-05T04:32:46Z","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/1908.05672/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"GPT-2 and BERT demonstrate the effectiveness of using pre-trained language models (LMs) on various natural language processing tasks. However, LM fine-tuning often suffers from catastrophic forgetting when applied to resource-rich tasks. In this work, we introduce a concerted training framework (CTNMT) that is the key to integrate the pre-trained LMs to neural machine translation (NMT). Our proposed CTNMT consists of three techniques: a) asymptotic distillation to ensure that the NMT model can retain the previous pre-trained knowledge; b) a dynamic switching gate to avoid catastrophic forgetti","authors_text":"Chengqi Zhao, Hao Zhou, Jiacheng Yang, Lei Li, Mingxuan Wang, Weinan Zhang, Yong Yu","cross_cats":["cs.LG"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2019-08-15T03:33:50Z","title":"Towards Making the Most of BERT in Neural Machine Translation"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1908.05672","kind":"arxiv","version":5},"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:b25ffa684c9ff834fb2d2273d83eb87b7f7ca3f157c901808fbced65278d385d","target":"record","created_at":"2026-07-05T04:32:46Z","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":"f3e7b313c4d2ef3ee12dcb5a0d0aa307649419e598262dc37d54724bcd8dea9d","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2019-08-15T03:33:50Z","title_canon_sha256":"14b3dc32f816ebccbf7f796d040affbbe568bf642a86d7b707e112238ea6761f"},"schema_version":"1.0","source":{"id":"1908.05672","kind":"arxiv","version":5}},"canonical_sha256":"fff97119ce239d409f1e28d5cefc28d4e139eb532c8cc42dc555e2216950f422","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"fff97119ce239d409f1e28d5cefc28d4e139eb532c8cc42dc555e2216950f422","first_computed_at":"2026-07-05T04:32:46.322934Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T04:32:46.322934Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"OaiwDyDkdLIdiJqMG+kfMpZDJH9pPmCdEm46ok8swRtJO5T7x5TyEwCNWsAmcU/c8OE9BMW/sAxUWDIyZhFZBg==","signature_status":"signed_v1","signed_at":"2026-07-05T04:32:46.323435Z","signed_message":"canonical_sha256_bytes"},"source_id":"1908.05672","source_kind":"arxiv","source_version":5}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:b25ffa684c9ff834fb2d2273d83eb87b7f7ca3f157c901808fbced65278d385d","sha256:92f88ad690f97f787ffb454a428a598056ab2c17a2e1a0278786864cb5467d05"],"state_sha256":"6e78ffa7542414ee33419c27e4b1e9f855ef954b858c929e88d05b7f94072a57"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"bKDdRcLFQsCz4coBYDRPMxy6B4sggxPRi1kH/6FWkIzc9h74tulCzs0IritAWkzuQwiBFrxpFajHQesk0YKjBw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-07T12:11:55.280609Z","bundle_sha256":"0fcdd22d9e03656472ff74556c106838d5dfa0f9de5742ab69eef0f284b41d85"}}