{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:BA552R42H7MKYMKONNGWZHQOJL","short_pith_number":"pith:BA552R42","schema_version":"1.0","canonical_sha256":"083bdd479a3fd8ac314e6b4d6c9e0e4ac53714ef24434a752ff8344b3d9b13a3","source":{"kind":"arxiv","id":"2307.15337","version":3},"attestation_state":"computed","paper":{"title":"Skeleton-of-Thought: Prompting LLMs for Efficient Parallel Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Huazhong Yang, Xuefei Ning, Yu Wang, Zifu Wang, Zinan Lin, Zixuan Zhou","submitted_at":"2023-07-28T06:31:34Z","abstract_excerpt":"This work aims at decreasing the end-to-end generation latency of large language models (LLMs). One of the major causes of the high generation latency is the sequential decoding approach adopted by almost all state-of-the-art LLMs. In this work, motivated by the thinking and writing process of humans, we propose Skeleton-of-Thought (SoT), which first guides LLMs to generate the skeleton of the answer, and then conducts parallel API calls or batched decoding to complete the contents of each skeleton point in parallel. Not only does SoT provide considerable speed-ups across 12 LLMs, but it can a"},"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":"2307.15337","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-07-28T06:31:34Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"f1197be0824a2fff85d90dd488da5cb00c4b2e864b3eb489771268185e573b02","abstract_canon_sha256":"66ea5b3146dfd391267b1f8fd402d72a604133e1137aee2654d3892b030dde90"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:51:09.962450Z","signature_b64":"5jVI9fWCjQDwxKLG6fwl12qQvnYY6NZNsPesLsmLAeUYwAD1X0GSXTnlH6glI76jQ1/ev1Lx1KuBiuX3DUA5Cg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"083bdd479a3fd8ac314e6b4d6c9e0e4ac53714ef24434a752ff8344b3d9b13a3","last_reissued_at":"2026-07-05T07:51:09.961991Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:51:09.961991Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Skeleton-of-Thought: Prompting LLMs for Efficient Parallel Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Huazhong Yang, Xuefei Ning, Yu Wang, Zifu Wang, Zinan Lin, Zixuan Zhou","submitted_at":"2023-07-28T06:31:34Z","abstract_excerpt":"This work aims at decreasing the end-to-end generation latency of large language models (LLMs). One of the major causes of the high generation latency is the sequential decoding approach adopted by almost all state-of-the-art LLMs. In this work, motivated by the thinking and writing process of humans, we propose Skeleton-of-Thought (SoT), which first guides LLMs to generate the skeleton of the answer, and then conducts parallel API calls or batched decoding to complete the contents of each skeleton point in parallel. Not only does SoT provide considerable speed-ups across 12 LLMs, but it can a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2307.15337","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/2307.15337/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":"2307.15337","created_at":"2026-07-05T07:51:09.962046+00:00"},{"alias_kind":"arxiv_version","alias_value":"2307.15337v3","created_at":"2026-07-05T07:51:09.962046+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2307.15337","created_at":"2026-07-05T07:51:09.962046+00:00"},{"alias_kind":"pith_short_12","alias_value":"BA552R42H7MK","created_at":"2026-07-05T07:51:09.962046+00:00"},{"alias_kind":"pith_short_16","alias_value":"BA552R42H7MKYMKO","created_at":"2026-07-05T07:51:09.962046+00:00"},{"alias_kind":"pith_short_8","alias_value":"BA552R42","created_at":"2026-07-05T07:51:09.962046+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":15,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.22385","citing_title":"MetaPS: Adaptive Programmatic Strategy Selection for Market Agents","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2606.07951","citing_title":"From `May' to `Is': Certainty Distortion in Language Model Rewriting","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01168","citing_title":"Thinking Economically: A Hierarchical Framework for Adaptive-Complexity Reasoning in LLMs","ref_index":56,"is_internal_anchor":false},{"citing_arxiv_id":"2504.02181","citing_title":"A Survey of Scaling in Large Language Model Reasoning","ref_index":144,"is_internal_anchor":false},{"citing_arxiv_id":"2507.21035","citing_title":"GenoMAS: A Multi-Agent Framework for Scientific Discovery via Code-Driven Gene Expression Analysis","ref_index":78,"is_internal_anchor":false},{"citing_arxiv_id":"2509.23322","citing_title":"Mitigating Visual Context Degradation in Large Multimodal Models: A Training-Free Decoupled Agentic Framework","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2412.13171","citing_title":"Compressed Chain of Thought: Efficient Reasoning Through Dense Representations","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2512.14044","citing_title":"OmniDrive-R1: Reinforcement-driven Interleaved Multi-modal Chain-of-Thought for Trustworthy Vision-Language Autonomous Driving","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2601.21619","citing_title":"On the Overscaling Curse of Parallel Thinking: System Efficacy Contradicts Sample Efficiency","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10195","citing_title":"Breaking the Reward Barrier: Accelerating Tree-of-Thought Reasoning via Speculative Exploration","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2404.14294","citing_title":"A Survey on Efficient Inference for Large Language Models","ref_index":47,"is_internal_anchor":false},{"citing_arxiv_id":"2309.01219","citing_title":"Siren's Song in the AI Ocean: A Survey on Hallucination in Large Language Models","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10195","citing_title":"Breaking the Reward Barrier: Accelerating Tree-of-Thought Reasoning via Speculative Exploration","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06914","citing_title":"Regulating Branch Parallelism in LLM Serving","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2604.20685","citing_title":"MGDA-Decoupled: Geometry-Aware Multi-Objective Optimisation for DPO-based LLM Alignment","ref_index":87,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/BA552R42H7MKYMKONNGWZHQOJL","json":"https://pith.science/pith/BA552R42H7MKYMKONNGWZHQOJL.json","graph_json":"https://pith.science/api/pith-number/BA552R42H7MKYMKONNGWZHQOJL/graph.json","events_json":"https://pith.science/api/pith-number/BA552R42H7MKYMKONNGWZHQOJL/events.json","paper":"https://pith.science/paper/BA552R42"},"agent_actions":{"view_html":"https://pith.science/pith/BA552R42H7MKYMKONNGWZHQOJL","download_json":"https://pith.science/pith/BA552R42H7MKYMKONNGWZHQOJL.json","view_paper":"https://pith.science/paper/BA552R42","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2307.15337&json=true","fetch_graph":"https://pith.science/api/pith-number/BA552R42H7MKYMKONNGWZHQOJL/graph.json","fetch_events":"https://pith.science/api/pith-number/BA552R42H7MKYMKONNGWZHQOJL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BA552R42H7MKYMKONNGWZHQOJL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BA552R42H7MKYMKONNGWZHQOJL/action/storage_attestation","attest_author":"https://pith.science/pith/BA552R42H7MKYMKONNGWZHQOJL/action/author_attestation","sign_citation":"https://pith.science/pith/BA552R42H7MKYMKONNGWZHQOJL/action/citation_signature","submit_replication":"https://pith.science/pith/BA552R42H7MKYMKONNGWZHQOJL/action/replication_record"}},"created_at":"2026-07-05T07:51:09.962046+00:00","updated_at":"2026-07-05T07:51:09.962046+00:00"}