{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:IK7MBVBOLHR7XX3Y5GJEPZSLH7","short_pith_number":"pith:IK7MBVBO","schema_version":"1.0","canonical_sha256":"42bec0d42e59e3fbdf78e99247e64b3fed7374045a8dfbb14e58722f80218f6e","source":{"kind":"arxiv","id":"2207.07061","version":2},"attestation_state":"computed","paper":{"title":"Confident Adaptive Language Modeling","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Adam Fisch, Dara Bahri, Donald Metzler, Jai Gupta, Mostafa Dehghani, Tal Schuster, Vinh Q. Tran, Yi Tay","submitted_at":"2022-07-14T17:00:19Z","abstract_excerpt":"Recent advances in Transformer-based large language models (LLMs) have led to significant performance improvements across many tasks. These gains come with a drastic increase in the models' size, potentially leading to slow and costly use at inference time. In practice, however, the series of generations made by LLMs is composed of varying levels of difficulty. While certain predictions truly benefit from the models' full capacity, other continuations are more trivial and can be solved with reduced compute. In this work, we introduce Confident Adaptive Language Modeling (CALM), a framework for"},"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":"2207.07061","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2022-07-14T17:00:19Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"306019b4bf7736525fa9f2f491d321e3d172b3715ca67cb385554f16a34d8bcb","abstract_canon_sha256":"e7a0cf7bfde076989d149a732f48e76aaacc7a1289da56ef772a4f62d3cb927b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:10:05.982981Z","signature_b64":"KK8jq6ulgJ+pghHhWKCFGTnB6vf7xqZHCF7doF5tPB0ekeVOSVDjaCojTLvsso3qrwyOKu0UWYwR7/QdFQryBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"42bec0d42e59e3fbdf78e99247e64b3fed7374045a8dfbb14e58722f80218f6e","last_reissued_at":"2026-07-05T05:10:05.982565Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:10:05.982565Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Confident Adaptive Language Modeling","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Adam Fisch, Dara Bahri, Donald Metzler, Jai Gupta, Mostafa Dehghani, Tal Schuster, Vinh Q. Tran, Yi Tay","submitted_at":"2022-07-14T17:00:19Z","abstract_excerpt":"Recent advances in Transformer-based large language models (LLMs) have led to significant performance improvements across many tasks. These gains come with a drastic increase in the models' size, potentially leading to slow and costly use at inference time. In practice, however, the series of generations made by LLMs is composed of varying levels of difficulty. While certain predictions truly benefit from the models' full capacity, other continuations are more trivial and can be solved with reduced compute. In this work, we introduce Confident Adaptive Language Modeling (CALM), a framework for"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2207.07061","kind":"arxiv","version":2},"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/2207.07061/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":"2207.07061","created_at":"2026-07-05T05:10:05.982621+00:00"},{"alias_kind":"arxiv_version","alias_value":"2207.07061v2","created_at":"2026-07-05T05:10:05.982621+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2207.07061","created_at":"2026-07-05T05:10:05.982621+00:00"},{"alias_kind":"pith_short_12","alias_value":"IK7MBVBOLHR7","created_at":"2026-07-05T05:10:05.982621+00:00"},{"alias_kind":"pith_short_16","alias_value":"IK7MBVBOLHR7XX3Y","created_at":"2026-07-05T05:10:05.982621+00:00"},{"alias_kind":"pith_short_8","alias_value":"IK7MBVBO","created_at":"2026-07-05T05:10:05.982621+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":7,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.23670","citing_title":"Tapered Language Models","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2606.09937","citing_title":"RKSC: Reasoning-Aware KV Cache Sharing and Confident Early Exit for Multi-Step LLM Inference","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09165","citing_title":"Sparse Layers are Critical to Scaling Looped Language Models","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2605.27712","citing_title":"Prefix-Safe Bayesian Belief Tracking for LLM Reasoning Reliability:Separating Calibration from Ranking","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2605.28919","citing_title":"CosmicFish-HRM: Adaptive Reasoning via Hierarchical Recurrent Mechanisms in Compact Language Models","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2303.08112","citing_title":"Eliciting Latent Predictions from Transformers with the Tuned Lens","ref_index":80,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09165","citing_title":"Sparse Layers are Critical to Scaling Looped Language Models","ref_index":10,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IK7MBVBOLHR7XX3Y5GJEPZSLH7","json":"https://pith.science/pith/IK7MBVBOLHR7XX3Y5GJEPZSLH7.json","graph_json":"https://pith.science/api/pith-number/IK7MBVBOLHR7XX3Y5GJEPZSLH7/graph.json","events_json":"https://pith.science/api/pith-number/IK7MBVBOLHR7XX3Y5GJEPZSLH7/events.json","paper":"https://pith.science/paper/IK7MBVBO"},"agent_actions":{"view_html":"https://pith.science/pith/IK7MBVBOLHR7XX3Y5GJEPZSLH7","download_json":"https://pith.science/pith/IK7MBVBOLHR7XX3Y5GJEPZSLH7.json","view_paper":"https://pith.science/paper/IK7MBVBO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2207.07061&json=true","fetch_graph":"https://pith.science/api/pith-number/IK7MBVBOLHR7XX3Y5GJEPZSLH7/graph.json","fetch_events":"https://pith.science/api/pith-number/IK7MBVBOLHR7XX3Y5GJEPZSLH7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IK7MBVBOLHR7XX3Y5GJEPZSLH7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IK7MBVBOLHR7XX3Y5GJEPZSLH7/action/storage_attestation","attest_author":"https://pith.science/pith/IK7MBVBOLHR7XX3Y5GJEPZSLH7/action/author_attestation","sign_citation":"https://pith.science/pith/IK7MBVBOLHR7XX3Y5GJEPZSLH7/action/citation_signature","submit_replication":"https://pith.science/pith/IK7MBVBOLHR7XX3Y5GJEPZSLH7/action/replication_record"}},"created_at":"2026-07-05T05:10:05.982621+00:00","updated_at":"2026-07-05T05:10:05.982621+00:00"}