{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:SFP26MYZI42C26TA7ZIRW6YTAV","short_pith_number":"pith:SFP26MYZ","schema_version":"1.0","canonical_sha256":"915faf331947342d7a60fe511b7b13057396e20ea949e76f9424f29fcee8819a","source":{"kind":"arxiv","id":"2410.07563","version":2},"attestation_state":"computed","paper":{"title":"PLaMo-100B: A Ground-Up Language Model Designed for Japanese Proficiency","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Daisuke Nishino, Daisuke Okanohara, Hiroaki Mikami, Hiroyoshi Komatsu, Kaizaburo Chubachi, Kentaro Imajo, Kosuke Nakago, Preferred Elements: Kenshin Abe, Shogo Murai, Shotaro Sano, Shuji Suzuki, Tianqi Xu, Toru Ogawa, Toshihiko Yanase, Toshiki Kataoka, Tsuguo Mogami, Yasuhiro Fujita, Yoshihiko Ozaki, Yuta Hirokawa","submitted_at":"2024-10-10T02:59:36Z","abstract_excerpt":"We introduce PLaMo-100B, a large-scale language model designed for Japanese proficiency. The model was trained from scratch using 2 trillion tokens, with architecture such as QK Normalization and Z-Loss to ensure training stability during the training process. Post-training techniques, including Supervised Fine-Tuning and Direct Preference Optimization, were applied to refine the model's performance. Benchmark evaluations suggest that PLaMo-100B performs well, particularly in Japanese-specific tasks, achieving results that are competitive with frontier models like GPT-4. The base model is avai"},"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":"2410.07563","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-10-10T02:59:36Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"7f5ccd94d36548b408d54755381ef47c14f204a361a2aececa8a2f83cb3a0c15","abstract_canon_sha256":"dc5b17d38d86a4c7b773dfbe92f9c34131ca157dc47fbf651f38098f994ad7a9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:24:00.185853Z","signature_b64":"v4NVBWERoguNg+vFEyUzuYyTA6hNx1K6bdMBdfxJlsGyKUKai6aOYquhxhqy5/Sy6Cqh9GQ9/2hTUpx8KphUDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"915faf331947342d7a60fe511b7b13057396e20ea949e76f9424f29fcee8819a","last_reissued_at":"2026-07-05T09:24:00.185321Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:24:00.185321Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"PLaMo-100B: A Ground-Up Language Model Designed for Japanese Proficiency","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Daisuke Nishino, Daisuke Okanohara, Hiroaki Mikami, Hiroyoshi Komatsu, Kaizaburo Chubachi, Kentaro Imajo, Kosuke Nakago, Preferred Elements: Kenshin Abe, Shogo Murai, Shotaro Sano, Shuji Suzuki, Tianqi Xu, Toru Ogawa, Toshihiko Yanase, Toshiki Kataoka, Tsuguo Mogami, Yasuhiro Fujita, Yoshihiko Ozaki, Yuta Hirokawa","submitted_at":"2024-10-10T02:59:36Z","abstract_excerpt":"We introduce PLaMo-100B, a large-scale language model designed for Japanese proficiency. The model was trained from scratch using 2 trillion tokens, with architecture such as QK Normalization and Z-Loss to ensure training stability during the training process. Post-training techniques, including Supervised Fine-Tuning and Direct Preference Optimization, were applied to refine the model's performance. Benchmark evaluations suggest that PLaMo-100B performs well, particularly in Japanese-specific tasks, achieving results that are competitive with frontier models like GPT-4. The base model is avai"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.07563","kind":"arxiv","version":2},"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/2410.07563/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":"2410.07563","created_at":"2026-07-05T09:24:00.185380+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.07563v2","created_at":"2026-07-05T09:24:00.185380+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.07563","created_at":"2026-07-05T09:24:00.185380+00:00"},{"alias_kind":"pith_short_12","alias_value":"SFP26MYZI42C","created_at":"2026-07-05T09:24:00.185380+00:00"},{"alias_kind":"pith_short_16","alias_value":"SFP26MYZI42C26TA","created_at":"2026-07-05T09:24:00.185380+00:00"},{"alias_kind":"pith_short_8","alias_value":"SFP26MYZ","created_at":"2026-07-05T09:24:00.185380+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2504.18080","citing_title":"Stabilizing Reasoning in Medical LLMs with Continued Pretraining and Reasoning Preference Optimization","ref_index":31,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/SFP26MYZI42C26TA7ZIRW6YTAV","json":"https://pith.science/pith/SFP26MYZI42C26TA7ZIRW6YTAV.json","graph_json":"https://pith.science/api/pith-number/SFP26MYZI42C26TA7ZIRW6YTAV/graph.json","events_json":"https://pith.science/api/pith-number/SFP26MYZI42C26TA7ZIRW6YTAV/events.json","paper":"https://pith.science/paper/SFP26MYZ"},"agent_actions":{"view_html":"https://pith.science/pith/SFP26MYZI42C26TA7ZIRW6YTAV","download_json":"https://pith.science/pith/SFP26MYZI42C26TA7ZIRW6YTAV.json","view_paper":"https://pith.science/paper/SFP26MYZ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.07563&json=true","fetch_graph":"https://pith.science/api/pith-number/SFP26MYZI42C26TA7ZIRW6YTAV/graph.json","fetch_events":"https://pith.science/api/pith-number/SFP26MYZI42C26TA7ZIRW6YTAV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/SFP26MYZI42C26TA7ZIRW6YTAV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/SFP26MYZI42C26TA7ZIRW6YTAV/action/storage_attestation","attest_author":"https://pith.science/pith/SFP26MYZI42C26TA7ZIRW6YTAV/action/author_attestation","sign_citation":"https://pith.science/pith/SFP26MYZI42C26TA7ZIRW6YTAV/action/citation_signature","submit_replication":"https://pith.science/pith/SFP26MYZI42C26TA7ZIRW6YTAV/action/replication_record"}},"created_at":"2026-07-05T09:24:00.185380+00:00","updated_at":"2026-07-05T09:24:00.185380+00:00"}