{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2021:NY4JKYXIQV3DC6IOW6CGFK5BXU","short_pith_number":"pith:NY4JKYXI","canonical_record":{"source":{"id":"2108.12284","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2021-08-26T17:26:56Z","cross_cats_sorted":["cs.AI","cs.NE"],"title_canon_sha256":"6147866b5534cf2b4ed3c745f0f1ede75de1627204521284572cd736787199cb","abstract_canon_sha256":"2b10d309603f3594fb604cf508e7bdcc8fa62934ef43d3f058de3529e225b33a"},"schema_version":"1.0"},"canonical_sha256":"6e389562e8857631790eb78462aba1bd3f37eee12c817f5b5ef1ea2d6c935a72","source":{"kind":"arxiv","id":"2108.12284","version":4},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2108.12284","created_at":"2026-07-05T03:56:30Z"},{"alias_kind":"arxiv_version","alias_value":"2108.12284v4","created_at":"2026-07-05T03:56:30Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2108.12284","created_at":"2026-07-05T03:56:30Z"},{"alias_kind":"pith_short_12","alias_value":"NY4JKYXIQV3D","created_at":"2026-07-05T03:56:30Z"},{"alias_kind":"pith_short_16","alias_value":"NY4JKYXIQV3DC6IO","created_at":"2026-07-05T03:56:30Z"},{"alias_kind":"pith_short_8","alias_value":"NY4JKYXI","created_at":"2026-07-05T03:56:30Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2021:NY4JKYXIQV3DC6IOW6CGFK5BXU","target":"record","payload":{"canonical_record":{"source":{"id":"2108.12284","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2021-08-26T17:26:56Z","cross_cats_sorted":["cs.AI","cs.NE"],"title_canon_sha256":"6147866b5534cf2b4ed3c745f0f1ede75de1627204521284572cd736787199cb","abstract_canon_sha256":"2b10d309603f3594fb604cf508e7bdcc8fa62934ef43d3f058de3529e225b33a"},"schema_version":"1.0"},"canonical_sha256":"6e389562e8857631790eb78462aba1bd3f37eee12c817f5b5ef1ea2d6c935a72","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:56:30.551028Z","signature_b64":"EbyKBtmHRYVz6Ss43X8FEQ7izCIeRRmwu28fpv4DMuu924clXOEOulIjIcx/7Ng2s1THLxpXSSp60ucJ4ylcDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6e389562e8857631790eb78462aba1bd3f37eee12c817f5b5ef1ea2d6c935a72","last_reissued_at":"2026-07-05T03:56:30.550562Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:56:30.550562Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2108.12284","source_version":4,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T03:56:30Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"WJK61+P9tUuAOhR5NE/0XxVJ5otHNvaRyHaUpg6OBw0wynVyboNIk9SO7645xGUvssm5Wj0PN+jEIllkRHH5Cg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-19T04:31:38.651062Z"},"content_sha256":"b2ffb93650b7a8552353b94c29a1b0ebea406c738e575e07b555b80414fb6b01","schema_version":"1.0","event_id":"sha256:b2ffb93650b7a8552353b94c29a1b0ebea406c738e575e07b555b80414fb6b01"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2021:NY4JKYXIQV3DC6IOW6CGFK5BXU","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"The Devil is in the Detail: Simple Tricks Improve Systematic Generalization of Transformers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.NE"],"primary_cat":"cs.LG","authors_text":"J\\\"urgen Schmidhuber, Kazuki Irie, R\\'obert Csord\\'as","submitted_at":"2021-08-26T17:26:56Z","abstract_excerpt":"Recently, many datasets have been proposed to test the systematic generalization ability of neural networks. The companion baseline Transformers, typically trained with default hyper-parameters from standard tasks, are shown to fail dramatically. Here we demonstrate that by revisiting model configurations as basic as scaling of embeddings, early stopping, relative positional embedding, and Universal Transformer variants, we can drastically improve the performance of Transformers on systematic generalization. We report improvements on five popular datasets: SCAN, CFQ, PCFG, COGS, and Mathematic"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2108.12284","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/2108.12284/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":"2026-07-05T03:56:30Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"ePha798rDGe/Zcv9eZL07biCPF+FHSvXsuIIxzTKf0HUEDmVRpjMwyJKh2LQOewntKCi/xYqnD3q3RkTtSASBQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-19T04:31:38.651664Z"},"content_sha256":"79dabc604f5d1880af279db780b89fe60609a6c3dea47380b59d3a7f67716980","schema_version":"1.0","event_id":"sha256:79dabc604f5d1880af279db780b89fe60609a6c3dea47380b59d3a7f67716980"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/NY4JKYXIQV3DC6IOW6CGFK5BXU/bundle.json","state_url":"https://pith.science/pith/NY4JKYXIQV3DC6IOW6CGFK5BXU/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/NY4JKYXIQV3DC6IOW6CGFK5BXU/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":"2026-08-19T04:31:38Z","links":{"resolver":"https://pith.science/pith/NY4JKYXIQV3DC6IOW6CGFK5BXU","bundle":"https://pith.science/pith/NY4JKYXIQV3DC6IOW6CGFK5BXU/bundle.json","state":"https://pith.science/pith/NY4JKYXIQV3DC6IOW6CGFK5BXU/state.json","well_known_bundle":"https://pith.science/.well-known/pith/NY4JKYXIQV3DC6IOW6CGFK5BXU/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2021:NY4JKYXIQV3DC6IOW6CGFK5BXU","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":"2b10d309603f3594fb604cf508e7bdcc8fa62934ef43d3f058de3529e225b33a","cross_cats_sorted":["cs.AI","cs.NE"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2021-08-26T17:26:56Z","title_canon_sha256":"6147866b5534cf2b4ed3c745f0f1ede75de1627204521284572cd736787199cb"},"schema_version":"1.0","source":{"id":"2108.12284","kind":"arxiv","version":4}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2108.12284","created_at":"2026-07-05T03:56:30Z"},{"alias_kind":"arxiv_version","alias_value":"2108.12284v4","created_at":"2026-07-05T03:56:30Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2108.12284","created_at":"2026-07-05T03:56:30Z"},{"alias_kind":"pith_short_12","alias_value":"NY4JKYXIQV3D","created_at":"2026-07-05T03:56:30Z"},{"alias_kind":"pith_short_16","alias_value":"NY4JKYXIQV3DC6IO","created_at":"2026-07-05T03:56:30Z"},{"alias_kind":"pith_short_8","alias_value":"NY4JKYXI","created_at":"2026-07-05T03:56:30Z"}],"graph_snapshots":[{"event_id":"sha256:79dabc604f5d1880af279db780b89fe60609a6c3dea47380b59d3a7f67716980","target":"graph","created_at":"2026-07-05T03:56:30Z","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/2108.12284/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Recently, many datasets have been proposed to test the systematic generalization ability of neural networks. The companion baseline Transformers, typically trained with default hyper-parameters from standard tasks, are shown to fail dramatically. Here we demonstrate that by revisiting model configurations as basic as scaling of embeddings, early stopping, relative positional embedding, and Universal Transformer variants, we can drastically improve the performance of Transformers on systematic generalization. We report improvements on five popular datasets: SCAN, CFQ, PCFG, COGS, and Mathematic","authors_text":"J\\\"urgen Schmidhuber, Kazuki Irie, R\\'obert Csord\\'as","cross_cats":["cs.AI","cs.NE"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2021-08-26T17:26:56Z","title":"The Devil is in the Detail: Simple Tricks Improve Systematic Generalization of Transformers"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2108.12284","kind":"arxiv","version":4},"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:b2ffb93650b7a8552353b94c29a1b0ebea406c738e575e07b555b80414fb6b01","target":"record","created_at":"2026-07-05T03:56:30Z","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":"2b10d309603f3594fb604cf508e7bdcc8fa62934ef43d3f058de3529e225b33a","cross_cats_sorted":["cs.AI","cs.NE"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2021-08-26T17:26:56Z","title_canon_sha256":"6147866b5534cf2b4ed3c745f0f1ede75de1627204521284572cd736787199cb"},"schema_version":"1.0","source":{"id":"2108.12284","kind":"arxiv","version":4}},"canonical_sha256":"6e389562e8857631790eb78462aba1bd3f37eee12c817f5b5ef1ea2d6c935a72","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"6e389562e8857631790eb78462aba1bd3f37eee12c817f5b5ef1ea2d6c935a72","first_computed_at":"2026-07-05T03:56:30.550562Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T03:56:30.550562Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"EbyKBtmHRYVz6Ss43X8FEQ7izCIeRRmwu28fpv4DMuu924clXOEOulIjIcx/7Ng2s1THLxpXSSp60ucJ4ylcDw==","signature_status":"signed_v1","signed_at":"2026-07-05T03:56:30.551028Z","signed_message":"canonical_sha256_bytes"},"source_id":"2108.12284","source_kind":"arxiv","source_version":4}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:b2ffb93650b7a8552353b94c29a1b0ebea406c738e575e07b555b80414fb6b01","sha256:79dabc604f5d1880af279db780b89fe60609a6c3dea47380b59d3a7f67716980"],"state_sha256":"8d0cf0d0ed2191adaa6bd0c85db1bdbfca2271de15e0a9ed43c9f4e20defb1a8"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"o6MGIHegdhGuZXR+tGBlYISwqAqxomrZwHGWS3gNHDdwXeK9Rk0zT6H1xk3JEtb3D5NrJwZDeKnPZvpCHSSLCw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-19T04:31:38.657341Z","bundle_sha256":"0ca5441cc8aeb4881be6561fabf83587fe5d5de70472cec7853ed8348fa95dcb"}}