{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:4AT4N2UU7VMCLJW4KVYUUQ6PLU","short_pith_number":"pith:4AT4N2UU","schema_version":"1.0","canonical_sha256":"e027c6ea94fd5825a6dc55714a43cf5d3633e991ed4150f3e4e812b81587049a","source":{"kind":"arxiv","id":"2606.28057","version":1},"attestation_state":"computed","paper":{"title":"MultiHashFormer: Hash-based Generative Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Atsuki Yamaguchi, Huiyin Xue, Nikolaos Aletras","submitted_at":"2026-06-26T13:03:29Z","abstract_excerpt":"Language models (LMs) represent tokens using embedding matrices that scale linearly with the vocabulary size. To constrain the parameter footprint, prior work proposes hashing many tokens into a single vector within encoder-only models. While this offers parameter efficiency, many-to-one collisions prevent its use in causal LMs. In this paper, we propose MultiHashFormer, a new framework that allows hash-based autoregression. Each token is represented as a unique hash signature, a short sequence of discrete hash IDs, generated by multiple independent hash functions. A Hash Encoder compresses th"},"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":"2606.28057","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-06-26T13:03:29Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"c769a5918c6e07f750d86ee4f186d569251ff31020c9b6d39ee51461ee49468a","abstract_canon_sha256":"c6d4d2d1d43104edc8d7d0fc696fb89aebbba0ebf2c72973043e9e78f83a33d9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-29T01:14:56.499229Z","signature_b64":"CHsZxMv1b5GSwcdJAqxw4NjQ4jaenasCL2bdpqFYBIVXDXzZO9SQQt8OCEvqWrqyRpbDzS52SM494ry2tiayAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e027c6ea94fd5825a6dc55714a43cf5d3633e991ed4150f3e4e812b81587049a","last_reissued_at":"2026-06-29T01:14:56.498804Z","signature_status":"signed_v1","first_computed_at":"2026-06-29T01:14:56.498804Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MultiHashFormer: Hash-based Generative Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Atsuki Yamaguchi, Huiyin Xue, Nikolaos Aletras","submitted_at":"2026-06-26T13:03:29Z","abstract_excerpt":"Language models (LMs) represent tokens using embedding matrices that scale linearly with the vocabulary size. To constrain the parameter footprint, prior work proposes hashing many tokens into a single vector within encoder-only models. While this offers parameter efficiency, many-to-one collisions prevent its use in causal LMs. In this paper, we propose MultiHashFormer, a new framework that allows hash-based autoregression. Each token is represented as a unique hash signature, a short sequence of discrete hash IDs, generated by multiple independent hash functions. A Hash Encoder compresses th"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.28057","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/2606.28057/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":"2606.28057","created_at":"2026-06-29T01:14:56.498859+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.28057v1","created_at":"2026-06-29T01:14:56.498859+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.28057","created_at":"2026-06-29T01:14:56.498859+00:00"},{"alias_kind":"pith_short_12","alias_value":"4AT4N2UU7VMC","created_at":"2026-06-29T01:14:56.498859+00:00"},{"alias_kind":"pith_short_16","alias_value":"4AT4N2UU7VMCLJW4","created_at":"2026-06-29T01:14:56.498859+00:00"},{"alias_kind":"pith_short_8","alias_value":"4AT4N2UU","created_at":"2026-06-29T01:14:56.498859+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4AT4N2UU7VMCLJW4KVYUUQ6PLU","json":"https://pith.science/pith/4AT4N2UU7VMCLJW4KVYUUQ6PLU.json","graph_json":"https://pith.science/api/pith-number/4AT4N2UU7VMCLJW4KVYUUQ6PLU/graph.json","events_json":"https://pith.science/api/pith-number/4AT4N2UU7VMCLJW4KVYUUQ6PLU/events.json","paper":"https://pith.science/paper/4AT4N2UU"},"agent_actions":{"view_html":"https://pith.science/pith/4AT4N2UU7VMCLJW4KVYUUQ6PLU","download_json":"https://pith.science/pith/4AT4N2UU7VMCLJW4KVYUUQ6PLU.json","view_paper":"https://pith.science/paper/4AT4N2UU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.28057&json=true","fetch_graph":"https://pith.science/api/pith-number/4AT4N2UU7VMCLJW4KVYUUQ6PLU/graph.json","fetch_events":"https://pith.science/api/pith-number/4AT4N2UU7VMCLJW4KVYUUQ6PLU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4AT4N2UU7VMCLJW4KVYUUQ6PLU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4AT4N2UU7VMCLJW4KVYUUQ6PLU/action/storage_attestation","attest_author":"https://pith.science/pith/4AT4N2UU7VMCLJW4KVYUUQ6PLU/action/author_attestation","sign_citation":"https://pith.science/pith/4AT4N2UU7VMCLJW4KVYUUQ6PLU/action/citation_signature","submit_replication":"https://pith.science/pith/4AT4N2UU7VMCLJW4KVYUUQ6PLU/action/replication_record"}},"created_at":"2026-06-29T01:14:56.498859+00:00","updated_at":"2026-06-29T01:14:56.498859+00:00"}