{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:QGYRHMFWMLYQOMPDKPPF6MVZ7N","short_pith_number":"pith:QGYRHMFW","canonical_record":{"source":{"id":"2304.09151","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-04-18T17:45:50Z","cross_cats_sorted":[],"title_canon_sha256":"291662e6585d81856dcdef8c8476eabb574805364481299715eee3a8bd691ac2","abstract_canon_sha256":"398ad3e5ee063a8d44cec92a8816fa6c2293ed314047283e934eb3f34aae03a8"},"schema_version":"1.0"},"canonical_sha256":"81b113b0b662f10731e353de5f32b9fb66d1b10a77216ba4f4dd021606cc16db","source":{"kind":"arxiv","id":"2304.09151","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2304.09151","created_at":"2026-07-05T06:02:17Z"},{"alias_kind":"arxiv_version","alias_value":"2304.09151v1","created_at":"2026-07-05T06:02:17Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2304.09151","created_at":"2026-07-05T06:02:17Z"},{"alias_kind":"pith_short_12","alias_value":"QGYRHMFWMLYQ","created_at":"2026-07-05T06:02:17Z"},{"alias_kind":"pith_short_16","alias_value":"QGYRHMFWMLYQOMPD","created_at":"2026-07-05T06:02:17Z"},{"alias_kind":"pith_short_8","alias_value":"QGYRHMFW","created_at":"2026-07-05T06:02:17Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:QGYRHMFWMLYQOMPDKPPF6MVZ7N","target":"record","payload":{"canonical_record":{"source":{"id":"2304.09151","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-04-18T17:45:50Z","cross_cats_sorted":[],"title_canon_sha256":"291662e6585d81856dcdef8c8476eabb574805364481299715eee3a8bd691ac2","abstract_canon_sha256":"398ad3e5ee063a8d44cec92a8816fa6c2293ed314047283e934eb3f34aae03a8"},"schema_version":"1.0"},"canonical_sha256":"81b113b0b662f10731e353de5f32b9fb66d1b10a77216ba4f4dd021606cc16db","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:02:17.674168Z","signature_b64":"YZGcZFrc/v3AfCMWozVbIpDCaQFraCbtbPh/ef3rLx2IPMfYNBe4oOEcJAN5hrDLSrhtTwjyyusgmPZ/Eza+BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"81b113b0b662f10731e353de5f32b9fb66d1b10a77216ba4f4dd021606cc16db","last_reissued_at":"2026-07-05T06:02:17.673553Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:02:17.673553Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2304.09151","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":"2026-07-05T06:02:17Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"b/mtZvlfq3UwxdDBeH6831Zs3RWwta+jcUrpzSsE10DYLWEFhgF0nbLmqBKs8AeKTJRUIJ7fHTXhBXyD4y22CQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-17T10:31:18.513152Z"},"content_sha256":"5e87f8eaef6de1289cd878542d3e2eb24a0cb3470d483db4a2484566b86c52ab","schema_version":"1.0","event_id":"sha256:5e87f8eaef6de1289cd878542d3e2eb24a0cb3470d483db4a2484566b86c52ab"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:QGYRHMFWMLYQOMPDKPPF6MVZ7N","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"UniMax: Fairer and more Effective Language Sampling for Large-Scale Multilingual Pretraining","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Adam Roberts, Hyung Won Chung, Noah Constant, Orhan Firat, Sharan Narang, Xavier Garcia, Yi Tay","submitted_at":"2023-04-18T17:45:50Z","abstract_excerpt":"Pretrained multilingual large language models have typically used heuristic temperature-based sampling to balance between different languages. However previous work has not systematically evaluated the efficacy of different pretraining language distributions across model scales. In this paper, we propose a new sampling method, UniMax, that delivers more uniform coverage of head languages while mitigating overfitting on tail languages by explicitly capping the number of repeats over each language's corpus. We perform an extensive series of ablations testing a range of sampling strategies on a s"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2304.09151","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/2304.09151/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":"2026-07-05T06:02:17Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"7sgBANoSZb+VkL2X2X9JbpzsuMRbSPCSvA0lYXzxngNjeNKMWWVcD9KNHo5CQDWbRzyFSRyetVNMeqKJ9xNODQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-17T10:31:18.513536Z"},"content_sha256":"9a43484ad00eb30c8af370ec74eb8f585d866ec077957113f3b731c00e66ec62","schema_version":"1.0","event_id":"sha256:9a43484ad00eb30c8af370ec74eb8f585d866ec077957113f3b731c00e66ec62"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/QGYRHMFWMLYQOMPDKPPF6MVZ7N/bundle.json","state_url":"https://pith.science/pith/QGYRHMFWMLYQOMPDKPPF6MVZ7N/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/QGYRHMFWMLYQOMPDKPPF6MVZ7N/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":"2026-07-17T10:31:18Z","links":{"resolver":"https://pith.science/pith/QGYRHMFWMLYQOMPDKPPF6MVZ7N","bundle":"https://pith.science/pith/QGYRHMFWMLYQOMPDKPPF6MVZ7N/bundle.json","state":"https://pith.science/pith/QGYRHMFWMLYQOMPDKPPF6MVZ7N/state.json","well_known_bundle":"https://pith.science/.well-known/pith/QGYRHMFWMLYQOMPDKPPF6MVZ7N/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:QGYRHMFWMLYQOMPDKPPF6MVZ7N","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":"398ad3e5ee063a8d44cec92a8816fa6c2293ed314047283e934eb3f34aae03a8","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-04-18T17:45:50Z","title_canon_sha256":"291662e6585d81856dcdef8c8476eabb574805364481299715eee3a8bd691ac2"},"schema_version":"1.0","source":{"id":"2304.09151","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2304.09151","created_at":"2026-07-05T06:02:17Z"},{"alias_kind":"arxiv_version","alias_value":"2304.09151v1","created_at":"2026-07-05T06:02:17Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2304.09151","created_at":"2026-07-05T06:02:17Z"},{"alias_kind":"pith_short_12","alias_value":"QGYRHMFWMLYQ","created_at":"2026-07-05T06:02:17Z"},{"alias_kind":"pith_short_16","alias_value":"QGYRHMFWMLYQOMPD","created_at":"2026-07-05T06:02:17Z"},{"alias_kind":"pith_short_8","alias_value":"QGYRHMFW","created_at":"2026-07-05T06:02:17Z"}],"graph_snapshots":[{"event_id":"sha256:9a43484ad00eb30c8af370ec74eb8f585d866ec077957113f3b731c00e66ec62","target":"graph","created_at":"2026-07-05T06:02:17Z","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/2304.09151/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Pretrained multilingual large language models have typically used heuristic temperature-based sampling to balance between different languages. However previous work has not systematically evaluated the efficacy of different pretraining language distributions across model scales. In this paper, we propose a new sampling method, UniMax, that delivers more uniform coverage of head languages while mitigating overfitting on tail languages by explicitly capping the number of repeats over each language's corpus. We perform an extensive series of ablations testing a range of sampling strategies on a s","authors_text":"Adam Roberts, Hyung Won Chung, Noah Constant, Orhan Firat, Sharan Narang, Xavier Garcia, Yi Tay","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-04-18T17:45:50Z","title":"UniMax: Fairer and more Effective Language Sampling for Large-Scale Multilingual Pretraining"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2304.09151","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":"sha256:5e87f8eaef6de1289cd878542d3e2eb24a0cb3470d483db4a2484566b86c52ab","target":"record","created_at":"2026-07-05T06:02:17Z","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":"398ad3e5ee063a8d44cec92a8816fa6c2293ed314047283e934eb3f34aae03a8","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-04-18T17:45:50Z","title_canon_sha256":"291662e6585d81856dcdef8c8476eabb574805364481299715eee3a8bd691ac2"},"schema_version":"1.0","source":{"id":"2304.09151","kind":"arxiv","version":1}},"canonical_sha256":"81b113b0b662f10731e353de5f32b9fb66d1b10a77216ba4f4dd021606cc16db","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"81b113b0b662f10731e353de5f32b9fb66d1b10a77216ba4f4dd021606cc16db","first_computed_at":"2026-07-05T06:02:17.673553Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T06:02:17.673553Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"YZGcZFrc/v3AfCMWozVbIpDCaQFraCbtbPh/ef3rLx2IPMfYNBe4oOEcJAN5hrDLSrhtTwjyyusgmPZ/Eza+BQ==","signature_status":"signed_v1","signed_at":"2026-07-05T06:02:17.674168Z","signed_message":"canonical_sha256_bytes"},"source_id":"2304.09151","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:5e87f8eaef6de1289cd878542d3e2eb24a0cb3470d483db4a2484566b86c52ab","sha256:9a43484ad00eb30c8af370ec74eb8f585d866ec077957113f3b731c00e66ec62"],"state_sha256":"e885e85ba0faa98a960790a5bb054c6cfdf8ea1dcf31506016f0cef642b567c8"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"gKi7LomYXoKifdSXVIS7DnJYDGcD/YBgB9T3uXabjMoiiW32y6q8Xu4pcmiCYXQkPud5iTrpEd+n/VvVs0guAw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-17T10:31:18.515585Z","bundle_sha256":"b6c9c1382bb39365cd0943c2a95e5cee5b310b665aac683db2165e9009090bc7"}}