{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:PMTRRNNA5VO5QX3BSI7TN7COYN","short_pith_number":"pith:PMTRRNNA","schema_version":"1.0","canonical_sha256":"7b2718b5a0ed5dd85f61923f36fc4ec340a742c55955b68440b2f5c5423b4773","source":{"kind":"arxiv","id":"2410.04707","version":1},"attestation_state":"computed","paper":{"title":"Learning How Hard to Think: Input-Adaptive Allocation of LM Computation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Andi Peng, Andreea Bobu, Idan Shenfeld, Jacob Andreas, Mehul Damani","submitted_at":"2024-10-07T02:52:30Z","abstract_excerpt":"Computationally intensive decoding procedures--including search, reranking, and self-critique--can improve the quality of language model (LM) outputs in problems spanning code generation, numerical reasoning, and dialog. Existing work typically applies the same decoding procedure for every input to an LM. But not all inputs require the same amount of computation to process. Can we allocate decoding computation adaptively, using more resources to answer questions whose answers will be harder to compute? We present an approach that predicts the distribution of rewards given an input and computat"},"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":"2410.04707","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-10-07T02:52:30Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"54235b4effcb53545fa0611ee0075595b665daeaba35b2cc48a66b42151f7a5d","abstract_canon_sha256":"a687f34c391e17808394baec5a04370523245a2380f4fb72e77df0e289f2460f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:16:54.953558Z","signature_b64":"VIMKtVrIUqGEQ5kVNTTgPIU93OSpBDWJEeVf7ACZ4zSO/bVya4WsavXmhWH0AAGMA8cZ7tY/3qzH4jnZvdxjBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7b2718b5a0ed5dd85f61923f36fc4ec340a742c55955b68440b2f5c5423b4773","last_reissued_at":"2026-07-05T09:16:54.953070Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:16:54.953070Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning How Hard to Think: Input-Adaptive Allocation of LM Computation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Andi Peng, Andreea Bobu, Idan Shenfeld, Jacob Andreas, Mehul Damani","submitted_at":"2024-10-07T02:52:30Z","abstract_excerpt":"Computationally intensive decoding procedures--including search, reranking, and self-critique--can improve the quality of language model (LM) outputs in problems spanning code generation, numerical reasoning, and dialog. Existing work typically applies the same decoding procedure for every input to an LM. But not all inputs require the same amount of computation to process. Can we allocate decoding computation adaptively, using more resources to answer questions whose answers will be harder to compute? We present an approach that predicts the distribution of rewards given an input and computat"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.04707","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/2410.04707/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":"2410.04707","created_at":"2026-07-05T09:16:54.953130+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.04707v1","created_at":"2026-07-05T09:16:54.953130+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.04707","created_at":"2026-07-05T09:16:54.953130+00:00"},{"alias_kind":"pith_short_12","alias_value":"PMTRRNNA5VO5","created_at":"2026-07-05T09:16:54.953130+00:00"},{"alias_kind":"pith_short_16","alias_value":"PMTRRNNA5VO5QX3B","created_at":"2026-07-05T09:16:54.953130+00:00"},{"alias_kind":"pith_short_8","alias_value":"PMTRRNNA","created_at":"2026-07-05T09:16:54.953130+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":9,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.01667","citing_title":"ATLAS: Agentic Test-time Learning-to-Allocate Scaling","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2606.31511","citing_title":"Falsification, Not Exposure: An Internally Preregistered Placebo-Controlled Decomposition of Self-Repair Feedback in Frozen Small Code Models","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15706","citing_title":"Differentiable Mixture-of-Agents Incentivizes Swarm Intelligence of Large Language Models","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2602.01970","citing_title":"Small Generalizable Prompt Predictive Models Can Steer Efficient RL Post-Training of Large Reasoning Models","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2602.16699","citing_title":"Calibrate-Then-Act: Cost-Aware Exploration in LLM Agents","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15706","citing_title":"Differentiable Mixture-of-Agents Incentivizes Swarm Intelligence of Large Language Models","ref_index":113,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17497","citing_title":"Self-Supervised On-Policy Distillation for Reasoning Language Models","ref_index":66,"is_internal_anchor":false},{"citing_arxiv_id":"2601.05110","citing_title":"GlimpRouter: Efficient Collaborative Inference by Glimpsing One Token of Thoughts","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2412.21187","citing_title":"Do NOT Think That Much for 2+3=? On the Overthinking of o1-Like LLMs","ref_index":3,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PMTRRNNA5VO5QX3BSI7TN7COYN","json":"https://pith.science/pith/PMTRRNNA5VO5QX3BSI7TN7COYN.json","graph_json":"https://pith.science/api/pith-number/PMTRRNNA5VO5QX3BSI7TN7COYN/graph.json","events_json":"https://pith.science/api/pith-number/PMTRRNNA5VO5QX3BSI7TN7COYN/events.json","paper":"https://pith.science/paper/PMTRRNNA"},"agent_actions":{"view_html":"https://pith.science/pith/PMTRRNNA5VO5QX3BSI7TN7COYN","download_json":"https://pith.science/pith/PMTRRNNA5VO5QX3BSI7TN7COYN.json","view_paper":"https://pith.science/paper/PMTRRNNA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.04707&json=true","fetch_graph":"https://pith.science/api/pith-number/PMTRRNNA5VO5QX3BSI7TN7COYN/graph.json","fetch_events":"https://pith.science/api/pith-number/PMTRRNNA5VO5QX3BSI7TN7COYN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PMTRRNNA5VO5QX3BSI7TN7COYN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PMTRRNNA5VO5QX3BSI7TN7COYN/action/storage_attestation","attest_author":"https://pith.science/pith/PMTRRNNA5VO5QX3BSI7TN7COYN/action/author_attestation","sign_citation":"https://pith.science/pith/PMTRRNNA5VO5QX3BSI7TN7COYN/action/citation_signature","submit_replication":"https://pith.science/pith/PMTRRNNA5VO5QX3BSI7TN7COYN/action/replication_record"}},"created_at":"2026-07-05T09:16:54.953130+00:00","updated_at":"2026-07-05T09:16:54.953130+00:00"}