{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:HBDKOTBBROD54AI4KMHJ6DPRKU","short_pith_number":"pith:HBDKOTBB","schema_version":"1.0","canonical_sha256":"3846a74c218b87de011c530e9f0df15503f90c6fdbf4441d6106f66da3d38750","source":{"kind":"arxiv","id":"2507.18391","version":1},"attestation_state":"computed","paper":{"title":"Revisiting LLM Reasoning via Information Bottleneck","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Dacheng Tao, Kai Jia, Shiye Lei, Zhihao Cheng","submitted_at":"2025-07-24T13:14:25Z","abstract_excerpt":"Large language models (LLMs) have recently demonstrated remarkable progress in reasoning capabilities through reinforcement learning with verifiable rewards (RLVR). By leveraging simple rule-based rewards, RL effectively incentivizes LLMs to produce extended chain-of-thought (CoT) reasoning trajectories, progressively guiding them toward correct answers. However, existing approaches remain largely heuristic and intuition-driven, limiting the development of principled methodologies. In this paper, we present a theoretical characterization of LLM reasoning grounded in information bottleneck (IB)"},"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":"2507.18391","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-07-24T13:14:25Z","cross_cats_sorted":[],"title_canon_sha256":"924ea3b81f065be82798bc0d18de84f10277f0c9b7266046c7251bd4e5af6d8e","abstract_canon_sha256":"7b484e2417b9b2b30e24ab21c6344ba3223f060b38391ca9125b7bfb016e9fc2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:42:47.327808Z","signature_b64":"Q4ErVAaDIuwOI+A94YpR6PmkWcThXRC8GYX/99gdxCxdh3NQ682M01AzYn1k0wYc7bZNBEPfYqLEsgevZR4GAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3846a74c218b87de011c530e9f0df15503f90c6fdbf4441d6106f66da3d38750","last_reissued_at":"2026-07-05T11:42:47.327161Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:42:47.327161Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Revisiting LLM Reasoning via Information Bottleneck","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Dacheng Tao, Kai Jia, Shiye Lei, Zhihao Cheng","submitted_at":"2025-07-24T13:14:25Z","abstract_excerpt":"Large language models (LLMs) have recently demonstrated remarkable progress in reasoning capabilities through reinforcement learning with verifiable rewards (RLVR). By leveraging simple rule-based rewards, RL effectively incentivizes LLMs to produce extended chain-of-thought (CoT) reasoning trajectories, progressively guiding them toward correct answers. However, existing approaches remain largely heuristic and intuition-driven, limiting the development of principled methodologies. In this paper, we present a theoretical characterization of LLM reasoning grounded in information bottleneck (IB)"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.18391","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/2507.18391/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":"2507.18391","created_at":"2026-07-05T11:42:47.327238+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.18391v1","created_at":"2026-07-05T11:42:47.327238+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.18391","created_at":"2026-07-05T11:42:47.327238+00:00"},{"alias_kind":"pith_short_12","alias_value":"HBDKOTBBROD5","created_at":"2026-07-05T11:42:47.327238+00:00"},{"alias_kind":"pith_short_16","alias_value":"HBDKOTBBROD54AI4","created_at":"2026-07-05T11:42:47.327238+00:00"},{"alias_kind":"pith_short_8","alias_value":"HBDKOTBB","created_at":"2026-07-05T11:42:47.327238+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2509.05489","citing_title":"Self-Aligned Reward: Towards Effective and Efficient Reasoners","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01111","citing_title":"When Less is Enough: Efficient Inference via Collaborative Reasoning","ref_index":23,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/HBDKOTBBROD54AI4KMHJ6DPRKU","json":"https://pith.science/pith/HBDKOTBBROD54AI4KMHJ6DPRKU.json","graph_json":"https://pith.science/api/pith-number/HBDKOTBBROD54AI4KMHJ6DPRKU/graph.json","events_json":"https://pith.science/api/pith-number/HBDKOTBBROD54AI4KMHJ6DPRKU/events.json","paper":"https://pith.science/paper/HBDKOTBB"},"agent_actions":{"view_html":"https://pith.science/pith/HBDKOTBBROD54AI4KMHJ6DPRKU","download_json":"https://pith.science/pith/HBDKOTBBROD54AI4KMHJ6DPRKU.json","view_paper":"https://pith.science/paper/HBDKOTBB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.18391&json=true","fetch_graph":"https://pith.science/api/pith-number/HBDKOTBBROD54AI4KMHJ6DPRKU/graph.json","fetch_events":"https://pith.science/api/pith-number/HBDKOTBBROD54AI4KMHJ6DPRKU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HBDKOTBBROD54AI4KMHJ6DPRKU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HBDKOTBBROD54AI4KMHJ6DPRKU/action/storage_attestation","attest_author":"https://pith.science/pith/HBDKOTBBROD54AI4KMHJ6DPRKU/action/author_attestation","sign_citation":"https://pith.science/pith/HBDKOTBBROD54AI4KMHJ6DPRKU/action/citation_signature","submit_replication":"https://pith.science/pith/HBDKOTBBROD54AI4KMHJ6DPRKU/action/replication_record"}},"created_at":"2026-07-05T11:42:47.327238+00:00","updated_at":"2026-07-05T11:42:47.327238+00:00"}