{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:E4NGIYJVUC5NCQZL2KTYPUFRNH","short_pith_number":"pith:E4NGIYJV","canonical_record":{"source":{"id":"2501.07314","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-01-13T13:26:50Z","cross_cats_sorted":[],"title_canon_sha256":"a4531a2aec421a693bfb5a6330c5d5a8e9aea6bf0ab27697af21228e528d6943","abstract_canon_sha256":"c0ac234c6f75a41f34819c6f33f124c44edbc01c8f03b45adfc8eb448c5b1376"},"schema_version":"1.0"},"canonical_sha256":"271a646135a0bad1432bd2a787d0b169ee8a56db1b98c97bf3a0463710afac5f","source":{"kind":"arxiv","id":"2501.07314","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2501.07314","created_at":"2026-07-05T10:00:18Z"},{"alias_kind":"arxiv_version","alias_value":"2501.07314v1","created_at":"2026-07-05T10:00:18Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.07314","created_at":"2026-07-05T10:00:18Z"},{"alias_kind":"pith_short_12","alias_value":"E4NGIYJVUC5N","created_at":"2026-07-05T10:00:18Z"},{"alias_kind":"pith_short_16","alias_value":"E4NGIYJVUC5NCQZL","created_at":"2026-07-05T10:00:18Z"},{"alias_kind":"pith_short_8","alias_value":"E4NGIYJV","created_at":"2026-07-05T10:00:18Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:E4NGIYJVUC5NCQZL2KTYPUFRNH","target":"record","payload":{"canonical_record":{"source":{"id":"2501.07314","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-01-13T13:26:50Z","cross_cats_sorted":[],"title_canon_sha256":"a4531a2aec421a693bfb5a6330c5d5a8e9aea6bf0ab27697af21228e528d6943","abstract_canon_sha256":"c0ac234c6f75a41f34819c6f33f124c44edbc01c8f03b45adfc8eb448c5b1376"},"schema_version":"1.0"},"canonical_sha256":"271a646135a0bad1432bd2a787d0b169ee8a56db1b98c97bf3a0463710afac5f","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:00:18.650612Z","signature_b64":"bLVaSy2l40tEyzjw4PKQyRdyANmPmEJ+aj7UqsmyCQvNIcjQlNH9o05ZooZ+AaAaLWsOS2dlYYKni8J7Cm1BBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"271a646135a0bad1432bd2a787d0b169ee8a56db1b98c97bf3a0463710afac5f","last_reissued_at":"2026-07-05T10:00:18.650124Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:00:18.650124Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2501.07314","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":"2026-07-05T10:00:18Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"pHAQoqG83DNYhmmXjsLv8L/DIh5j9I4X3a2plL7LPEhgSVY4n4aelShGJGZmDh7VEkZAWHjfr8em3ZKhDF2TDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-12T17:51:22.589392Z"},"content_sha256":"98f4b44d9257524ec118f5b81db9e5828273f05ecf2a9c0960b93eb0745b4371","schema_version":"1.0","event_id":"sha256:98f4b44d9257524ec118f5b81db9e5828273f05ecf2a9c0960b93eb0745b4371"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:E4NGIYJVUC5NCQZL2KTYPUFRNH","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"FinerWeb-10BT: Refining Web Data with LLM-Based Line-Level Filtering","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Erik Henriksson, Filip Ginter, Otto Tarkka","submitted_at":"2025-01-13T13:26:50Z","abstract_excerpt":"Data quality is crucial for training Large Language Models (LLMs). Traditional heuristic filters often miss low-quality text or mistakenly remove valuable content. In this paper, we introduce an LLM-based line-level filtering method to enhance training data quality. We use GPT-4o mini to label a 20,000-document sample from FineWeb at the line level, allowing the model to create descriptive labels for low-quality lines. These labels are grouped into nine main categories, and we train a DeBERTa-v3 classifier to scale the filtering to a 10B-token subset of FineWeb. To test the impact of our filte"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.07314","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/2501.07314/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":"2026-07-05T10:00:18Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"PUkDS+xuWBsp7x9Sp1kpWUmlhBRlXXQhH4utWQqxotxvKn0CX455M6nDLDpyAf4IHdfebal3LLA6+zulEjmzDw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-12T17:51:22.589892Z"},"content_sha256":"cf2084680e87ef1e6a26e0376d02e33fb322a16be76079c8d0bdf4112a5a30bf","schema_version":"1.0","event_id":"sha256:cf2084680e87ef1e6a26e0376d02e33fb322a16be76079c8d0bdf4112a5a30bf"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/E4NGIYJVUC5NCQZL2KTYPUFRNH/bundle.json","state_url":"https://pith.science/pith/E4NGIYJVUC5NCQZL2KTYPUFRNH/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/E4NGIYJVUC5NCQZL2KTYPUFRNH/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":"2026-08-12T17:51:22Z","links":{"resolver":"https://pith.science/pith/E4NGIYJVUC5NCQZL2KTYPUFRNH","bundle":"https://pith.science/pith/E4NGIYJVUC5NCQZL2KTYPUFRNH/bundle.json","state":"https://pith.science/pith/E4NGIYJVUC5NCQZL2KTYPUFRNH/state.json","well_known_bundle":"https://pith.science/.well-known/pith/E4NGIYJVUC5NCQZL2KTYPUFRNH/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:E4NGIYJVUC5NCQZL2KTYPUFRNH","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":"c0ac234c6f75a41f34819c6f33f124c44edbc01c8f03b45adfc8eb448c5b1376","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-01-13T13:26:50Z","title_canon_sha256":"a4531a2aec421a693bfb5a6330c5d5a8e9aea6bf0ab27697af21228e528d6943"},"schema_version":"1.0","source":{"id":"2501.07314","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2501.07314","created_at":"2026-07-05T10:00:18Z"},{"alias_kind":"arxiv_version","alias_value":"2501.07314v1","created_at":"2026-07-05T10:00:18Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.07314","created_at":"2026-07-05T10:00:18Z"},{"alias_kind":"pith_short_12","alias_value":"E4NGIYJVUC5N","created_at":"2026-07-05T10:00:18Z"},{"alias_kind":"pith_short_16","alias_value":"E4NGIYJVUC5NCQZL","created_at":"2026-07-05T10:00:18Z"},{"alias_kind":"pith_short_8","alias_value":"E4NGIYJV","created_at":"2026-07-05T10:00:18Z"}],"graph_snapshots":[{"event_id":"sha256:cf2084680e87ef1e6a26e0376d02e33fb322a16be76079c8d0bdf4112a5a30bf","target":"graph","created_at":"2026-07-05T10:00:18Z","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/2501.07314/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Data quality is crucial for training Large Language Models (LLMs). Traditional heuristic filters often miss low-quality text or mistakenly remove valuable content. In this paper, we introduce an LLM-based line-level filtering method to enhance training data quality. We use GPT-4o mini to label a 20,000-document sample from FineWeb at the line level, allowing the model to create descriptive labels for low-quality lines. These labels are grouped into nine main categories, and we train a DeBERTa-v3 classifier to scale the filtering to a 10B-token subset of FineWeb. To test the impact of our filte","authors_text":"Erik Henriksson, Filip Ginter, Otto Tarkka","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-01-13T13:26:50Z","title":"FinerWeb-10BT: Refining Web Data with LLM-Based Line-Level Filtering"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.07314","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":"sha256:98f4b44d9257524ec118f5b81db9e5828273f05ecf2a9c0960b93eb0745b4371","target":"record","created_at":"2026-07-05T10:00:18Z","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":"c0ac234c6f75a41f34819c6f33f124c44edbc01c8f03b45adfc8eb448c5b1376","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-01-13T13:26:50Z","title_canon_sha256":"a4531a2aec421a693bfb5a6330c5d5a8e9aea6bf0ab27697af21228e528d6943"},"schema_version":"1.0","source":{"id":"2501.07314","kind":"arxiv","version":1}},"canonical_sha256":"271a646135a0bad1432bd2a787d0b169ee8a56db1b98c97bf3a0463710afac5f","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"271a646135a0bad1432bd2a787d0b169ee8a56db1b98c97bf3a0463710afac5f","first_computed_at":"2026-07-05T10:00:18.650124Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:00:18.650124Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"bLVaSy2l40tEyzjw4PKQyRdyANmPmEJ+aj7UqsmyCQvNIcjQlNH9o05ZooZ+AaAaLWsOS2dlYYKni8J7Cm1BBw==","signature_status":"signed_v1","signed_at":"2026-07-05T10:00:18.650612Z","signed_message":"canonical_sha256_bytes"},"source_id":"2501.07314","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:98f4b44d9257524ec118f5b81db9e5828273f05ecf2a9c0960b93eb0745b4371","sha256:cf2084680e87ef1e6a26e0376d02e33fb322a16be76079c8d0bdf4112a5a30bf"],"state_sha256":"5f29cd94548ce73e08055251287ca12eab4ee8629ce719e62194b3d5197aed27"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"goVhcfC6fH0H8f/Zem8ffxEwdNrCv4rCXTzy9Wv9817cc1jJ6axl1pZY9ELtyMfqqqLrDW0RLf8Q2NrBMi8yAA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-12T17:51:22.595767Z","bundle_sha256":"f5d8bb5e0bb6e90ab2cc36c5103923f3a6742c232d6e98f359c4d3b9852cbea7"}}