{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:OJFRJBQUELOD6KEXSX2E6DE622","short_pith_number":"pith:OJFRJBQU","canonical_record":{"source":{"id":"2505.15634","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-05-21T15:17:59Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"cb7876d287be8925888cc5c2c23f6bcced5b56e77fb2d832bfe59072b2ad2921","abstract_canon_sha256":"36d60cc0a3a41909b68269cb3bb4799b2ceff3b0492965eff57651e3790d43d8"},"schema_version":"1.0"},"canonical_sha256":"724b14861422dc3f289795f44f0c9ed6b2a9904d5387d9da89f63dbd564d3e5d","source":{"kind":"arxiv","id":"2505.15634","version":4},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2505.15634","created_at":"2026-07-05T11:35:56Z"},{"alias_kind":"arxiv_version","alias_value":"2505.15634v4","created_at":"2026-07-05T11:35:56Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.15634","created_at":"2026-07-05T11:35:56Z"},{"alias_kind":"pith_short_12","alias_value":"OJFRJBQUELOD","created_at":"2026-07-05T11:35:56Z"},{"alias_kind":"pith_short_16","alias_value":"OJFRJBQUELOD6KEX","created_at":"2026-07-05T11:35:56Z"},{"alias_kind":"pith_short_8","alias_value":"OJFRJBQU","created_at":"2026-07-05T11:35:56Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:OJFRJBQUELOD6KEXSX2E6DE622","target":"record","payload":{"canonical_record":{"source":{"id":"2505.15634","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-05-21T15:17:59Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"cb7876d287be8925888cc5c2c23f6bcced5b56e77fb2d832bfe59072b2ad2921","abstract_canon_sha256":"36d60cc0a3a41909b68269cb3bb4799b2ceff3b0492965eff57651e3790d43d8"},"schema_version":"1.0"},"canonical_sha256":"724b14861422dc3f289795f44f0c9ed6b2a9904d5387d9da89f63dbd564d3e5d","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:35:56.070763Z","signature_b64":"c8wVdLtWlIimtaKEukQzqcqYTUJ+ZPY7tlwS2EYIfY16NYfmZykV8JT39a41upKZqXB1vwklwzwaYn86fGSlAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"724b14861422dc3f289795f44f0c9ed6b2a9904d5387d9da89f63dbd564d3e5d","last_reissued_at":"2026-07-05T11:35:56.070231Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:35:56.070231Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2505.15634","source_version":4,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T11:35:56Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"kOOVJGzbGSy/8jBMu493oqqiZWSgtK/M0q/hTGigT57pdZzcTG6uKYurkiywb1IwipAznXOhhOFH15/l31IfBg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-10T05:00:23.696937Z"},"content_sha256":"5ec870656a9b34ba5384fbdd3cd65fc86c39cf872efdfc73983cf04060139f99","schema_version":"1.0","event_id":"sha256:5ec870656a9b34ba5384fbdd3cd65fc86c39cf872efdfc73983cf04060139f99"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:OJFRJBQUELOD6KEXSX2E6DE622","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Feature Extraction and Steering for Enhanced Chain-of-Thought Reasoning in Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Haoyi Xiong, Mengnan Du, Xu Wang, Yuzhe Yang, Zihao Li, Ziyu Yao","submitted_at":"2025-05-21T15:17:59Z","abstract_excerpt":"Large Language Models (LLMs) demonstrate the ability to solve reasoning and mathematical problems using the Chain-of-Thought (CoT) technique. Expanding CoT length, as seen in models such as DeepSeek-R1, significantly enhances this reasoning for complex problems, but requires costly and high-quality long CoT data and fine-tuning. This work, inspired by the deep thinking paradigm of DeepSeek-R1, utilizes a steering technique to enhance the reasoning ability of an LLM without external datasets. Our method first employs Sparse Autoencoders (SAEs) to extract interpretable features from vanilla CoT."},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.15634","kind":"arxiv","version":4},"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/2505.15634/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T11:35:56Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"LTpnr7V0BPS4nZPCb187llvsO0HDwRkrXFCuI5rh6Wd02i2WivnsyZ970gR93pHNP2jgdTnL0GkJTjAC9qHeBA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-10T05:00:23.697847Z"},"content_sha256":"324eb44a8eea19eca5c69bb56f9f1f346f7b7b71b02bad17d56bccdc8c06afe5","schema_version":"1.0","event_id":"sha256:324eb44a8eea19eca5c69bb56f9f1f346f7b7b71b02bad17d56bccdc8c06afe5"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/OJFRJBQUELOD6KEXSX2E6DE622/bundle.json","state_url":"https://pith.science/pith/OJFRJBQUELOD6KEXSX2E6DE622/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/OJFRJBQUELOD6KEXSX2E6DE622/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-10T05:00:23Z","links":{"resolver":"https://pith.science/pith/OJFRJBQUELOD6KEXSX2E6DE622","bundle":"https://pith.science/pith/OJFRJBQUELOD6KEXSX2E6DE622/bundle.json","state":"https://pith.science/pith/OJFRJBQUELOD6KEXSX2E6DE622/state.json","well_known_bundle":"https://pith.science/.well-known/pith/OJFRJBQUELOD6KEXSX2E6DE622/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:OJFRJBQUELOD6KEXSX2E6DE622","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"36d60cc0a3a41909b68269cb3bb4799b2ceff3b0492965eff57651e3790d43d8","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-05-21T15:17:59Z","title_canon_sha256":"cb7876d287be8925888cc5c2c23f6bcced5b56e77fb2d832bfe59072b2ad2921"},"schema_version":"1.0","source":{"id":"2505.15634","kind":"arxiv","version":4}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2505.15634","created_at":"2026-07-05T11:35:56Z"},{"alias_kind":"arxiv_version","alias_value":"2505.15634v4","created_at":"2026-07-05T11:35:56Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.15634","created_at":"2026-07-05T11:35:56Z"},{"alias_kind":"pith_short_12","alias_value":"OJFRJBQUELOD","created_at":"2026-07-05T11:35:56Z"},{"alias_kind":"pith_short_16","alias_value":"OJFRJBQUELOD6KEX","created_at":"2026-07-05T11:35:56Z"},{"alias_kind":"pith_short_8","alias_value":"OJFRJBQU","created_at":"2026-07-05T11:35:56Z"}],"graph_snapshots":[{"event_id":"sha256:324eb44a8eea19eca5c69bb56f9f1f346f7b7b71b02bad17d56bccdc8c06afe5","target":"graph","created_at":"2026-07-05T11:35:56Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2505.15634/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large Language Models (LLMs) demonstrate the ability to solve reasoning and mathematical problems using the Chain-of-Thought (CoT) technique. Expanding CoT length, as seen in models such as DeepSeek-R1, significantly enhances this reasoning for complex problems, but requires costly and high-quality long CoT data and fine-tuning. This work, inspired by the deep thinking paradigm of DeepSeek-R1, utilizes a steering technique to enhance the reasoning ability of an LLM without external datasets. Our method first employs Sparse Autoencoders (SAEs) to extract interpretable features from vanilla CoT.","authors_text":"Haoyi Xiong, Mengnan Du, Xu Wang, Yuzhe Yang, Zihao Li, Ziyu Yao","cross_cats":["cs.LG"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-05-21T15:17:59Z","title":"Feature Extraction and Steering for Enhanced Chain-of-Thought Reasoning in Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.15634","kind":"arxiv","version":4},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:5ec870656a9b34ba5384fbdd3cd65fc86c39cf872efdfc73983cf04060139f99","target":"record","created_at":"2026-07-05T11:35:56Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"36d60cc0a3a41909b68269cb3bb4799b2ceff3b0492965eff57651e3790d43d8","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-05-21T15:17:59Z","title_canon_sha256":"cb7876d287be8925888cc5c2c23f6bcced5b56e77fb2d832bfe59072b2ad2921"},"schema_version":"1.0","source":{"id":"2505.15634","kind":"arxiv","version":4}},"canonical_sha256":"724b14861422dc3f289795f44f0c9ed6b2a9904d5387d9da89f63dbd564d3e5d","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"724b14861422dc3f289795f44f0c9ed6b2a9904d5387d9da89f63dbd564d3e5d","first_computed_at":"2026-07-05T11:35:56.070231Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:35:56.070231Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"c8wVdLtWlIimtaKEukQzqcqYTUJ+ZPY7tlwS2EYIfY16NYfmZykV8JT39a41upKZqXB1vwklwzwaYn86fGSlAA==","signature_status":"signed_v1","signed_at":"2026-07-05T11:35:56.070763Z","signed_message":"canonical_sha256_bytes"},"source_id":"2505.15634","source_kind":"arxiv","source_version":4}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:5ec870656a9b34ba5384fbdd3cd65fc86c39cf872efdfc73983cf04060139f99","sha256:324eb44a8eea19eca5c69bb56f9f1f346f7b7b71b02bad17d56bccdc8c06afe5"],"state_sha256":"f9f6fc8c65d1a7c28abb73c49bc81bdf96304ea8aba47c6617a5f32b8405618c"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"WNOW5koYAWXbo4tzmRvxM4qzqRQaOI0FGp/uoyrKX3BJp0bWAdZs+suqG1gfK5hTmOIeKNRw+IrtQctu+LBwDw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-10T05:00:23.728873Z","bundle_sha256":"bf821a392957cde02071e56512cd53ad29a71ecdfcec33fb9ec1d69a2760d9fe"}}