{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:TTOR5WYFI2IYTAON2KS732KX6R","short_pith_number":"pith:TTOR5WYF","schema_version":"1.0","canonical_sha256":"9cdd1edb0546918981cdd2a5fde957f46df40b9b45943936b0b15fa07ff3345f","source":{"kind":"arxiv","id":"2310.07923","version":5},"attestation_state":"computed","paper":{"title":"The Expressive Power of Transformers with Chain of Thought","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CC","cs.CL","cs.LO"],"primary_cat":"cs.LG","authors_text":"Ashish Sabharwal, William Merrill","submitted_at":"2023-10-11T22:35:18Z","abstract_excerpt":"Recent theoretical work has identified surprisingly simple reasoning problems, such as checking if two nodes in a graph are connected or simulating finite-state machines, that are provably unsolvable by standard transformers that answer immediately after reading their input. However, in practice, transformers' reasoning can be improved by allowing them to use a \"chain of thought\" or \"scratchpad\", i.e., generate and condition on a sequence of intermediate tokens before answering. Motivated by this, we ask: Does such intermediate generation fundamentally extend the computational power of a decod"},"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":"2310.07923","kind":"arxiv","version":5},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-10-11T22:35:18Z","cross_cats_sorted":["cs.CC","cs.CL","cs.LO"],"title_canon_sha256":"94a5f121cb2264403515843dc13e37244dc9efa5c172fb1c1f19def6418c75e0","abstract_canon_sha256":"45d9488a024be91ec58c476cbf02c60e7b4e4aeb75260e5a7c0854186b78b122"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:07:00.603340Z","signature_b64":"Dddyoh0WhEyp+UoFaA/tmXvNiJ7Ozjw7jZ/2/bn0Blcl648K4Z+K62tGAV58qBlv/EbiRLNHF2xlQL28RoypAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9cdd1edb0546918981cdd2a5fde957f46df40b9b45943936b0b15fa07ff3345f","last_reissued_at":"2026-07-05T08:07:00.602880Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:07:00.602880Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The Expressive Power of Transformers with Chain of Thought","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CC","cs.CL","cs.LO"],"primary_cat":"cs.LG","authors_text":"Ashish Sabharwal, William Merrill","submitted_at":"2023-10-11T22:35:18Z","abstract_excerpt":"Recent theoretical work has identified surprisingly simple reasoning problems, such as checking if two nodes in a graph are connected or simulating finite-state machines, that are provably unsolvable by standard transformers that answer immediately after reading their input. However, in practice, transformers' reasoning can be improved by allowing them to use a \"chain of thought\" or \"scratchpad\", i.e., generate and condition on a sequence of intermediate tokens before answering. Motivated by this, we ask: Does such intermediate generation fundamentally extend the computational power of a decod"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.07923","kind":"arxiv","version":5},"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/2310.07923/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":"2310.07923","created_at":"2026-07-05T08:07:00.602937+00:00"},{"alias_kind":"arxiv_version","alias_value":"2310.07923v5","created_at":"2026-07-05T08:07:00.602937+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.07923","created_at":"2026-07-05T08:07:00.602937+00:00"},{"alias_kind":"pith_short_12","alias_value":"TTOR5WYFI2IY","created_at":"2026-07-05T08:07:00.602937+00:00"},{"alias_kind":"pith_short_16","alias_value":"TTOR5WYFI2IYTAON","created_at":"2026-07-05T08:07:00.602937+00:00"},{"alias_kind":"pith_short_8","alias_value":"TTOR5WYF","created_at":"2026-07-05T08:07:00.602937+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":27,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2607.06720","citing_title":"When Does In-Context Search Help? A Sampling-Complexity Theory of Reflection-Driven Reasoning","ref_index":11,"is_internal_anchor":true},{"citing_arxiv_id":"2607.07423","citing_title":"The Optimal Sample Complexity of Learning Autoregressive Chain-of-Thought","ref_index":8,"is_internal_anchor":true},{"citing_arxiv_id":"2606.26969","citing_title":"Einstein World Models","ref_index":87,"is_internal_anchor":false},{"citing_arxiv_id":"2606.18089","citing_title":"From Reasoning Traces to Reusable Modules: Understanding Compositional Generalization in Language Model Reasoning","ref_index":184,"is_internal_anchor":false},{"citing_arxiv_id":"2606.09731","citing_title":"Tight Sample Complexity of Transformers","ref_index":43,"is_internal_anchor":false},{"citing_arxiv_id":"2606.06479","citing_title":"Pretraining Recurrent Networks without Recurrence","ref_index":84,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01532","citing_title":"Rethinking the Role of Positional Encoding: Sliding-Window Transformers without PE Remain Turing Complete","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2605.20531","citing_title":"Pseudo-Formalization for Automatic Proof Verification","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2605.28600","citing_title":"Transformers Provably Learn to Internalize Chain-of-Thought","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00183","citing_title":"Agentic Transformers Provably Learn to Search via Reinforcement Learning","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2501.19201","citing_title":"Efficient Reasoning with Hidden Thinking","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2602.08167","citing_title":"Self-Supervised Bootstrapping of Action-Predictive Embodied Reasoning","ref_index":48,"is_internal_anchor":false},{"citing_arxiv_id":"2605.20531","citing_title":"Pseudo-Formalization for Automatic Proof Verification","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2507.12549","citing_title":"The Serial Scaling Hypothesis","ref_index":70,"is_internal_anchor":false},{"citing_arxiv_id":"2508.16745","citing_title":"Beyond Memorization: Extending Reasoning Depth with Recurrence, Memory and Test-Time Compute Scaling","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2604.02371","citing_title":"Internalized Reasoning for Long-Context Visual Document Understanding","ref_index":34,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08221","citing_title":"NoisyCoconut: Counterfactual Consensus via Latent Space Reasoning","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2604.22951","citing_title":"The Power of Power Law: Asymmetry Enables Compositional Reasoning","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2412.06769","citing_title":"Training Large Language Models to Reason in a Continuous Latent Space","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2604.12426","citing_title":"Do Transformers Use their Depth Adaptively? Evidence from a Relational Reasoning Task","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2604.11535","citing_title":"Problem Reductions at Scale: Agentic Integration of Computationally Hard Problems","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2604.06799","citing_title":"Beyond Accuracy: Diagnosing Algebraic Reasoning Failures in LLMs Across Nine Complexity Dimensions","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06819","citing_title":"A Theory of Online Learning with Autoregressive Chain-of-Thought Reasoning","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2403.07974","citing_title":"LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Code","ref_index":96,"is_internal_anchor":false},{"citing_arxiv_id":"2604.14727","citing_title":"Geometric Capacity of Transformers: A Tropical Geometry Perspective","ref_index":20,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/TTOR5WYFI2IYTAON2KS732KX6R","json":"https://pith.science/pith/TTOR5WYFI2IYTAON2KS732KX6R.json","graph_json":"https://pith.science/api/pith-number/TTOR5WYFI2IYTAON2KS732KX6R/graph.json","events_json":"https://pith.science/api/pith-number/TTOR5WYFI2IYTAON2KS732KX6R/events.json","paper":"https://pith.science/paper/TTOR5WYF"},"agent_actions":{"view_html":"https://pith.science/pith/TTOR5WYFI2IYTAON2KS732KX6R","download_json":"https://pith.science/pith/TTOR5WYFI2IYTAON2KS732KX6R.json","view_paper":"https://pith.science/paper/TTOR5WYF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2310.07923&json=true","fetch_graph":"https://pith.science/api/pith-number/TTOR5WYFI2IYTAON2KS732KX6R/graph.json","fetch_events":"https://pith.science/api/pith-number/TTOR5WYFI2IYTAON2KS732KX6R/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TTOR5WYFI2IYTAON2KS732KX6R/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TTOR5WYFI2IYTAON2KS732KX6R/action/storage_attestation","attest_author":"https://pith.science/pith/TTOR5WYFI2IYTAON2KS732KX6R/action/author_attestation","sign_citation":"https://pith.science/pith/TTOR5WYFI2IYTAON2KS732KX6R/action/citation_signature","submit_replication":"https://pith.science/pith/TTOR5WYFI2IYTAON2KS732KX6R/action/replication_record"}},"created_at":"2026-07-05T08:07:00.602937+00:00","updated_at":"2026-07-05T08:07:00.602937+00:00"}