{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:CJRRFPPAGPN2N254GP2TLVE6V6","short_pith_number":"pith:CJRRFPPA","schema_version":"1.0","canonical_sha256":"126312bde033dba6ebbc33f535d49eafa8d47d4a7c708e5bcf6c9acbe6830425","source":{"kind":"arxiv","id":"2210.13393","version":1},"attestation_state":"computed","paper":{"title":"We need to talk about random seeds","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Steven Bethard","submitted_at":"2022-10-24T16:48:45Z","abstract_excerpt":"Modern neural network libraries all take as a hyperparameter a random seed, typically used to determine the initial state of the model parameters. This opinion piece argues that there are some safe uses for random seeds: as part of the hyperparameter search to select a good model, creating an ensemble of several models, or measuring the sensitivity of the training algorithm to the random seed hyperparameter. It argues that some uses for random seeds are risky: using a fixed random seed for \"replicability\" and varying only the random seed to create score distributions for performance comparison"},"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":"2210.13393","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-10-24T16:48:45Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"5257e177c4d251013b02b0ba4800ee9bb7acfe8ce65fd43a392b63b1beaf1944","abstract_canon_sha256":"913e36ce5ddea62cc228d69c6c99c1821e6eae0a98b07ae0ffb38659a8726278"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:09:44.992355Z","signature_b64":"9U7cusNC0xlSb0SALSwAuw4cceaQBS58qgWDiHC6ij+disPw991KWFdCmCPrO+p+y/xMYgP+sZaxTPDq9Z5ECA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"126312bde033dba6ebbc33f535d49eafa8d47d4a7c708e5bcf6c9acbe6830425","last_reissued_at":"2026-07-05T05:09:44.991899Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:09:44.991899Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"We need to talk about random seeds","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Steven Bethard","submitted_at":"2022-10-24T16:48:45Z","abstract_excerpt":"Modern neural network libraries all take as a hyperparameter a random seed, typically used to determine the initial state of the model parameters. This opinion piece argues that there are some safe uses for random seeds: as part of the hyperparameter search to select a good model, creating an ensemble of several models, or measuring the sensitivity of the training algorithm to the random seed hyperparameter. It argues that some uses for random seeds are risky: using a fixed random seed for \"replicability\" and varying only the random seed to create score distributions for performance comparison"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2210.13393","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/2210.13393/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":"2210.13393","created_at":"2026-07-05T05:09:44.991956+00:00"},{"alias_kind":"arxiv_version","alias_value":"2210.13393v1","created_at":"2026-07-05T05:09:44.991956+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2210.13393","created_at":"2026-07-05T05:09:44.991956+00:00"},{"alias_kind":"pith_short_12","alias_value":"CJRRFPPAGPN2","created_at":"2026-07-05T05:09:44.991956+00:00"},{"alias_kind":"pith_short_16","alias_value":"CJRRFPPAGPN2N254","created_at":"2026-07-05T05:09:44.991956+00:00"},{"alias_kind":"pith_short_8","alias_value":"CJRRFPPA","created_at":"2026-07-05T05:09:44.991956+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.22917","citing_title":"GRAIN: Group Aggregation via Min-Norm Objective","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2606.22053","citing_title":"Gradient-Descent Steps to Success over Mean Accuracy: A Paradigm Shift for ML","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11206","citing_title":"Instructions Shape Production of Language, not Processing","ref_index":197,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11206","citing_title":"Instructions Shape Production of Language, not Processing","ref_index":197,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/CJRRFPPAGPN2N254GP2TLVE6V6","json":"https://pith.science/pith/CJRRFPPAGPN2N254GP2TLVE6V6.json","graph_json":"https://pith.science/api/pith-number/CJRRFPPAGPN2N254GP2TLVE6V6/graph.json","events_json":"https://pith.science/api/pith-number/CJRRFPPAGPN2N254GP2TLVE6V6/events.json","paper":"https://pith.science/paper/CJRRFPPA"},"agent_actions":{"view_html":"https://pith.science/pith/CJRRFPPAGPN2N254GP2TLVE6V6","download_json":"https://pith.science/pith/CJRRFPPAGPN2N254GP2TLVE6V6.json","view_paper":"https://pith.science/paper/CJRRFPPA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2210.13393&json=true","fetch_graph":"https://pith.science/api/pith-number/CJRRFPPAGPN2N254GP2TLVE6V6/graph.json","fetch_events":"https://pith.science/api/pith-number/CJRRFPPAGPN2N254GP2TLVE6V6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CJRRFPPAGPN2N254GP2TLVE6V6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CJRRFPPAGPN2N254GP2TLVE6V6/action/storage_attestation","attest_author":"https://pith.science/pith/CJRRFPPAGPN2N254GP2TLVE6V6/action/author_attestation","sign_citation":"https://pith.science/pith/CJRRFPPAGPN2N254GP2TLVE6V6/action/citation_signature","submit_replication":"https://pith.science/pith/CJRRFPPAGPN2N254GP2TLVE6V6/action/replication_record"}},"created_at":"2026-07-05T05:09:44.991956+00:00","updated_at":"2026-07-05T05:09:44.991956+00:00"}