{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:AWEE722BT4DEKB27SPHVYJCN2T","short_pith_number":"pith:AWEE722B","schema_version":"1.0","canonical_sha256":"05884feb419f0645075f93cf5c244dd4e6f0feaa47dbfb7b849926b51552621f","source":{"kind":"arxiv","id":"2601.06199","version":3},"attestation_state":"computed","paper":{"title":"FastSLM: Hierarchical Temporal Abstraction for Efficient Long-Form Speech Adaptation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.SD"],"primary_cat":"eess.AS","authors_text":"Chang-Jae Chun, Junseok Lee, Sangyong Lee","submitted_at":"2026-01-08T07:46:03Z","abstract_excerpt":"Scaling Multimodal Large Language Models (MLLMs) to long-form speech is bottlenecked by the explosive growth of input tokens. Unlike images or videos, audio lacks overlapping information, making extreme 1-token compression highly susceptible to the loss of fine-grained acoustic cues. To overcome this, we propose FastSLM, a token-efficient architecture featuring the Hierarchical Temporal Abstractor (HTA). HTA progressively distills non-overlapping acoustic features across multiple temporal scales, achieving an extreme compression rate of 1.67 tokens per second a 97% reduction without losing cri"},"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":"2601.06199","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"eess.AS","submitted_at":"2026-01-08T07:46:03Z","cross_cats_sorted":["cs.AI","cs.SD"],"title_canon_sha256":"3e50087d8dd8a36810ad941395d3839f30e6fa862071fcf2c843c26b46d97848","abstract_canon_sha256":"800cd2e474040816c7a664d68f408d450d00ec6d7f707e77c4204eb8ba3578f3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T02:04:12.955112Z","signature_b64":"g3b0CmfDM9xPv/YNqjve4c/huMPfNJO62dqHmUjXr08A2EDyAFI6PaGcfao+aX/ejLkFOehx47w/D8HPeIxLCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"05884feb419f0645075f93cf5c244dd4e6f0feaa47dbfb7b849926b51552621f","last_reissued_at":"2026-06-02T02:04:12.954636Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T02:04:12.954636Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"FastSLM: Hierarchical Temporal Abstraction for Efficient Long-Form Speech Adaptation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.SD"],"primary_cat":"eess.AS","authors_text":"Chang-Jae Chun, Junseok Lee, Sangyong Lee","submitted_at":"2026-01-08T07:46:03Z","abstract_excerpt":"Scaling Multimodal Large Language Models (MLLMs) to long-form speech is bottlenecked by the explosive growth of input tokens. Unlike images or videos, audio lacks overlapping information, making extreme 1-token compression highly susceptible to the loss of fine-grained acoustic cues. To overcome this, we propose FastSLM, a token-efficient architecture featuring the Hierarchical Temporal Abstractor (HTA). HTA progressively distills non-overlapping acoustic features across multiple temporal scales, achieving an extreme compression rate of 1.67 tokens per second a 97% reduction without losing cri"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2601.06199","kind":"arxiv","version":3},"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/2601.06199/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":"2601.06199","created_at":"2026-06-02T02:04:12.954699+00:00"},{"alias_kind":"arxiv_version","alias_value":"2601.06199v3","created_at":"2026-06-02T02:04:12.954699+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2601.06199","created_at":"2026-06-02T02:04:12.954699+00:00"},{"alias_kind":"pith_short_12","alias_value":"AWEE722BT4DE","created_at":"2026-06-02T02:04:12.954699+00:00"},{"alias_kind":"pith_short_16","alias_value":"AWEE722BT4DEKB27","created_at":"2026-06-02T02:04:12.954699+00:00"},{"alias_kind":"pith_short_8","alias_value":"AWEE722B","created_at":"2026-06-02T02:04:12.954699+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2604.22245","citing_title":"Listening with Time: Precise Temporal Awareness for Long-Form Audio Understanding","ref_index":21,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/AWEE722BT4DEKB27SPHVYJCN2T","json":"https://pith.science/pith/AWEE722BT4DEKB27SPHVYJCN2T.json","graph_json":"https://pith.science/api/pith-number/AWEE722BT4DEKB27SPHVYJCN2T/graph.json","events_json":"https://pith.science/api/pith-number/AWEE722BT4DEKB27SPHVYJCN2T/events.json","paper":"https://pith.science/paper/AWEE722B"},"agent_actions":{"view_html":"https://pith.science/pith/AWEE722BT4DEKB27SPHVYJCN2T","download_json":"https://pith.science/pith/AWEE722BT4DEKB27SPHVYJCN2T.json","view_paper":"https://pith.science/paper/AWEE722B","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2601.06199&json=true","fetch_graph":"https://pith.science/api/pith-number/AWEE722BT4DEKB27SPHVYJCN2T/graph.json","fetch_events":"https://pith.science/api/pith-number/AWEE722BT4DEKB27SPHVYJCN2T/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AWEE722BT4DEKB27SPHVYJCN2T/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AWEE722BT4DEKB27SPHVYJCN2T/action/storage_attestation","attest_author":"https://pith.science/pith/AWEE722BT4DEKB27SPHVYJCN2T/action/author_attestation","sign_citation":"https://pith.science/pith/AWEE722BT4DEKB27SPHVYJCN2T/action/citation_signature","submit_replication":"https://pith.science/pith/AWEE722BT4DEKB27SPHVYJCN2T/action/replication_record"}},"created_at":"2026-06-02T02:04:12.954699+00:00","updated_at":"2026-06-02T02:04:12.954699+00:00"}