{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:ZNS2TPASXW6KQPCR423XQD4ZQH","short_pith_number":"pith:ZNS2TPAS","schema_version":"1.0","canonical_sha256":"cb65a9bc12bdbca83c51e6b7780f9981ce353ba0311298a9a5d09454008d62c5","source":{"kind":"arxiv","id":"2310.05424","version":1},"attestation_state":"computed","paper":{"title":"Fast and Robust Early-Exiting Framework for Autoregressive Language Models with Synchronized Parallel Decoding","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Hwanjun Song, Jongwoo Ko, Sangmin Bae, Se-Young Yun","submitted_at":"2023-10-09T05:53:05Z","abstract_excerpt":"To tackle the high inference latency exhibited by autoregressive language models, previous studies have proposed an early-exiting framework that allocates adaptive computation paths for each token based on the complexity of generating the subsequent token. However, we observed several shortcomings, including performance degradation caused by a state copying mechanism or numerous exit paths, and sensitivity to exit confidence thresholds. Consequently, we propose a Fast and Robust Early-Exiting (FREE) framework, which incorporates a shallow-deep module and a synchronized parallel decoding. Our f"},"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":"2310.05424","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-10-09T05:53:05Z","cross_cats_sorted":[],"title_canon_sha256":"4610544fdb928b93e539000ab4954e41dec767d0fdd05346535d8476efc9e6d0","abstract_canon_sha256":"5a15ca60904815cf9d685606974dcf5cc9761b8d82998ad29c7d0caeb574db9a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:58:45.122404Z","signature_b64":"y2G6+rfdrksTFmPunRwXjjDQuBxLdXT9J/lvgNT3HWKlOXJMzEK98BUDycc/3OfhNxB0paTjhqyoVF6li2grBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cb65a9bc12bdbca83c51e6b7780f9981ce353ba0311298a9a5d09454008d62c5","last_reissued_at":"2026-07-05T06:58:45.121881Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:58:45.121881Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Fast and Robust Early-Exiting Framework for Autoregressive Language Models with Synchronized Parallel Decoding","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Hwanjun Song, Jongwoo Ko, Sangmin Bae, Se-Young Yun","submitted_at":"2023-10-09T05:53:05Z","abstract_excerpt":"To tackle the high inference latency exhibited by autoregressive language models, previous studies have proposed an early-exiting framework that allocates adaptive computation paths for each token based on the complexity of generating the subsequent token. However, we observed several shortcomings, including performance degradation caused by a state copying mechanism or numerous exit paths, and sensitivity to exit confidence thresholds. Consequently, we propose a Fast and Robust Early-Exiting (FREE) framework, which incorporates a shallow-deep module and a synchronized parallel decoding. Our f"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.05424","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/2310.05424/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":"2310.05424","created_at":"2026-07-05T06:58:45.121938+00:00"},{"alias_kind":"arxiv_version","alias_value":"2310.05424v1","created_at":"2026-07-05T06:58:45.121938+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.05424","created_at":"2026-07-05T06:58:45.121938+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZNS2TPASXW6K","created_at":"2026-07-05T06:58:45.121938+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZNS2TPASXW6KQPCR","created_at":"2026-07-05T06:58:45.121938+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZNS2TPAS","created_at":"2026-07-05T06:58:45.121938+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2507.01449","citing_title":"LogitSpec: Accelerating Retrieval-based Speculative Decoding via Next Next Token Speculation","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2604.09607","citing_title":"LLM-assisted Agentic Edge Intelligence Framework","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2604.20503","citing_title":"FASER: Fine-Grained Phase Management for Speculative Decoding in Dynamic LLM Serving","ref_index":4,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZNS2TPASXW6KQPCR423XQD4ZQH","json":"https://pith.science/pith/ZNS2TPASXW6KQPCR423XQD4ZQH.json","graph_json":"https://pith.science/api/pith-number/ZNS2TPASXW6KQPCR423XQD4ZQH/graph.json","events_json":"https://pith.science/api/pith-number/ZNS2TPASXW6KQPCR423XQD4ZQH/events.json","paper":"https://pith.science/paper/ZNS2TPAS"},"agent_actions":{"view_html":"https://pith.science/pith/ZNS2TPASXW6KQPCR423XQD4ZQH","download_json":"https://pith.science/pith/ZNS2TPASXW6KQPCR423XQD4ZQH.json","view_paper":"https://pith.science/paper/ZNS2TPAS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2310.05424&json=true","fetch_graph":"https://pith.science/api/pith-number/ZNS2TPASXW6KQPCR423XQD4ZQH/graph.json","fetch_events":"https://pith.science/api/pith-number/ZNS2TPASXW6KQPCR423XQD4ZQH/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZNS2TPASXW6KQPCR423XQD4ZQH/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZNS2TPASXW6KQPCR423XQD4ZQH/action/storage_attestation","attest_author":"https://pith.science/pith/ZNS2TPASXW6KQPCR423XQD4ZQH/action/author_attestation","sign_citation":"https://pith.science/pith/ZNS2TPASXW6KQPCR423XQD4ZQH/action/citation_signature","submit_replication":"https://pith.science/pith/ZNS2TPASXW6KQPCR423XQD4ZQH/action/replication_record"}},"created_at":"2026-07-05T06:58:45.121938+00:00","updated_at":"2026-07-05T06:58:45.121938+00:00"}