{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2021:L7OYSFSVBQGOGCT5BUTOA7RHKQ","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":"d78706899687440daa186035fd2e634b3d5d6e2e1b75f396c7f0c8126dc10670","cross_cats_sorted":["cs.CC","cs.IT","cs.LG","math.IT"],"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2021-06-29T04:37:23Z","title_canon_sha256":"a35ac4ef54910d0d4db133514dbe866a811a9c7012d8c391786118426128934b"},"schema_version":"1.0","source":{"id":"2107.02027","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2107.02027","created_at":"2026-07-05T05:03:49Z"},{"alias_kind":"arxiv_version","alias_value":"2107.02027v2","created_at":"2026-07-05T05:03:49Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2107.02027","created_at":"2026-07-05T05:03:49Z"},{"alias_kind":"pith_short_12","alias_value":"L7OYSFSVBQGO","created_at":"2026-07-05T05:03:49Z"},{"alias_kind":"pith_short_16","alias_value":"L7OYSFSVBQGOGCT5","created_at":"2026-07-05T05:03:49Z"},{"alias_kind":"pith_short_8","alias_value":"L7OYSFSV","created_at":"2026-07-05T05:03:49Z"}],"graph_snapshots":[{"event_id":"sha256:49ff52b5bcaa889615fc3af3559586a7ab7fb9fb3bd4b917a8927bfc98515f9c","target":"graph","created_at":"2026-07-05T05:03:49Z","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/2107.02027/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Effective training of today's large language models (LLMs) depends on large batches and long sequences for throughput and accuracy. To handle variable-length sequences on hardware accelerators, it is common practice to introduce padding tokens, so that all sequences in a batch have the same length. We show in this paper that the variation in sequence lengths in common NLP datasets is such that up to 50% of all tokens can be padding. In less common, but not extreme, cases (e.g. GLUE-cola with sequence length 128), the ratio is up to 89%. Existing methods to address the resulting inefficiency ar","authors_text":"Andrew Fitzgibbon, Mario Michael Krell, Matej Kosec, Sergio P. Perez","cross_cats":["cs.CC","cs.IT","cs.LG","math.IT"],"headline":"","license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2021-06-29T04:37:23Z","title":"Efficient Sequence Packing without Cross-contamination: Accelerating Large Language Models without Impacting Performance"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2107.02027","kind":"arxiv","version":2},"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:f3949f97b13698a77551e83fb9dcfdb6625229dbb63b03431ac3a3d7851cfa65","target":"record","created_at":"2026-07-05T05:03:49Z","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":"d78706899687440daa186035fd2e634b3d5d6e2e1b75f396c7f0c8126dc10670","cross_cats_sorted":["cs.CC","cs.IT","cs.LG","math.IT"],"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2021-06-29T04:37:23Z","title_canon_sha256":"a35ac4ef54910d0d4db133514dbe866a811a9c7012d8c391786118426128934b"},"schema_version":"1.0","source":{"id":"2107.02027","kind":"arxiv","version":2}},"canonical_sha256":"5fdd8916550c0ce30a7d0d26e07e27543421862204fbafbedae2fc0771ebfc72","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"5fdd8916550c0ce30a7d0d26e07e27543421862204fbafbedae2fc0771ebfc72","first_computed_at":"2026-07-05T05:03:49.410131Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T05:03:49.410131Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"m0e8BfHOcJ6YwJns0HAWaEAmK6fbkvQ/Q15764PiNxuwXPCUYornM3fWLPo9X59cnbdAOaXUDmyfR3r3yooOAQ==","signature_status":"signed_v1","signed_at":"2026-07-05T05:03:49.410615Z","signed_message":"canonical_sha256_bytes"},"source_id":"2107.02027","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:f3949f97b13698a77551e83fb9dcfdb6625229dbb63b03431ac3a3d7851cfa65","sha256:49ff52b5bcaa889615fc3af3559586a7ab7fb9fb3bd4b917a8927bfc98515f9c"],"state_sha256":"acd714ba95b48c6ee3fb2998743f424d9e678a792ccf496883e2c97540cb0a5a"}