{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:ECALI4OFANEXYE2OSNPJB2AP4N","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":"5f018de7b629e31317483c2263c951ff3f6f73fd51bd0d2334ccf054c10150e2","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.SE","submitted_at":"2026-05-13T20:09:35Z","title_canon_sha256":"6aba564456d26590cf6e7e2caa1c38f0c4de2add3d160cc377eeb28ed81ddd11"},"schema_version":"1.0","source":{"id":"2605.14084","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.14084","created_at":"2026-05-17T23:39:12Z"},{"alias_kind":"arxiv_version","alias_value":"2605.14084v1","created_at":"2026-05-17T23:39:12Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.14084","created_at":"2026-05-17T23:39:12Z"},{"alias_kind":"pith_short_12","alias_value":"ECALI4OFANEX","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"ECALI4OFANEXYE2O","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"ECALI4OF","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:68d006d379838f136152a67665dcdfa6ecefb2aecf1e23648e698a480429b540","target":"graph","created_at":"2026-05-17T23:39:12Z","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":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"By merging paired Instruct and Thinking checkpoints, CRANE delivers strong gains over either individual model while preserving Instruct-level efficiency: on Roo-Eval it achieves pass1 of 66.2% (+19.5%) for Qwen3-30B-A3B and 81.5% (+8.7%) for Qwen3-Next-80B-A3B; on SWE-bench-Verified it resolves up to 14 additional instances at both scales (122/500 and 180/500); and on Terminal-Bench v2 it improves pass1/pass5 by up to 2.3%/7.8%, reaching 7.6%/17.9% and 14.8%/30.3%, respectively, consistently outperforming alternative merging strategies across all three benchmarks."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That the Thinking-Instruct delta vector contains directional information that can be selectively transferred via magnitude thresholding, Conservative Taylor Gate, and Graduated Sigmoidal Projection without degrading tool-use protocols or introducing new failure modes."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"CRANE merges Instruct and Thinking model checkpoints via constrained nullspace editing to improve code agent reasoning and benchmark performance without retraining."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"CRANE injects selected reasoning directions from thinking checkpoints into instruct models by editing their parameter nullspace, raising code agent success rates on benchmarks while keeping tool-use efficiency."}],"snapshot_sha256":"616447d40d0cd1ec1642b09cc23906aca6d39df32be06afc62f5732e36c6ad77"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"paper":{"abstract_excerpt":"Code agents must both reason over long-horizon repository state and obey strict tool-use protocols. In paired Instruct/Thinking checkpoints, these capabilities are complementary but misaligned. The Instruct model is concise and tool-disciplined, whereas the Thinking model offers stronger planning and recovery behavior but often over-deliberates and degrades agent performance. We present CRANE (Constrained Reasoning Injection for Code Agents via Nullspace Editing), a training-free parameter-editing method that treats the Thinking-Instruct delta as a directional pool of candidate reasoning edits","authors_text":"Michele Merler, Mingzhi Zhu, Raju Pavuluri, Stacy Patterson","cross_cats":["cs.AI","cs.CL"],"headline":"CRANE injects selected reasoning directions from thinking checkpoints into instruct models by editing their parameter nullspace, raising code agent success rates on benchmarks while keeping tool-use efficiency.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.SE","submitted_at":"2026-05-13T20:09:35Z","title":"CRANE: Constrained Reasoning Injection for Code Agents via Nullspace Editing"},"references":{"count":21,"internal_anchors":7,"resolved_work":21,"sample":[{"cited_arxiv_id":"2603.00729","doi":"","is_internal_anchor":true,"ref_index":1,"title":"Qwen3-Coder-Next Technical Report","work_id":"ad966e68-641d-4b33-a9da-57cf741f35a6","year":2026},{"cited_arxiv_id":"2501.12948","doi":"","is_internal_anchor":true,"ref_index":2,"title":"DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning","work_id":"e6b75ad5-2877-4168-97c8-710407094d20","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Lorahub: Efficient cross-task generalization via dynamic lora composition","work_id":"66ee831a-2b33-423d-87ec-ba52b341a21b","year":null},{"cited_arxiv_id":"2310.06770","doi":"","is_internal_anchor":true,"ref_index":4,"title":"SWE-bench: Can Language Models Resolve Real-World GitHub Issues?","work_id":"d0effe15-a689-441a-8e3f-ea35f1c4e4b1","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"arXiv preprint arXiv:2505.22113 , year=","work_id":"1eb7cd22-ed7f-48ef-be86-59775a274675","year":null}],"snapshot_sha256":"8e50bd1b8a5fe7bf2a0bf23fcc1a9212cdf947d8adc8cc87d7fdd9996534443c"},"source":{"id":"2605.14084","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-15T04:42:27.537662Z","id":"df310626-5be2-45b9-8a4e-e48893b0fea6","model_set":{"reader":"grok-4.3"},"one_line_summary":"CRANE merges Instruct and Thinking model checkpoints via constrained nullspace editing to improve code agent reasoning and benchmark performance without retraining.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"CRANE injects selected reasoning directions from thinking checkpoints into instruct models by editing their parameter nullspace, raising code agent success rates on benchmarks while keeping tool-use efficiency.","strongest_claim":"By merging paired Instruct and Thinking checkpoints, CRANE delivers strong gains over either individual model while preserving Instruct-level efficiency: on Roo-Eval it achieves pass1 of 66.2% (+19.5%) for Qwen3-30B-A3B and 81.5% (+8.7%) for Qwen3-Next-80B-A3B; on SWE-bench-Verified it resolves up to 14 additional instances at both scales (122/500 and 180/500); and on Terminal-Bench v2 it improves pass1/pass5 by up to 2.3%/7.8%, reaching 7.6%/17.9% and 14.8%/30.3%, respectively, consistently outperforming alternative merging strategies across all three benchmarks.","weakest_assumption":"That the Thinking-Instruct delta vector contains directional information that can be selectively transferred via magnitude thresholding, Conservative Taylor Gate, and Graduated Sigmoidal Projection without degrading tool-use protocols or introducing new failure modes."}},"verdict_id":"df310626-5be2-45b9-8a4e-e48893b0fea6"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:f8525f18b5db78224940a564491b33867f3eba4734a0e385597055f9f1343472","target":"record","created_at":"2026-05-17T23:39:12Z","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":"5f018de7b629e31317483c2263c951ff3f6f73fd51bd0d2334ccf054c10150e2","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.SE","submitted_at":"2026-05-13T20:09:35Z","title_canon_sha256":"6aba564456d26590cf6e7e2caa1c38f0c4de2add3d160cc377eeb28ed81ddd11"},"schema_version":"1.0","source":{"id":"2605.14084","kind":"arxiv","version":1}},"canonical_sha256":"2080b471c503497c134e935e90e80fe3507589c75eabd5d1635af5fc842f3213","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"2080b471c503497c134e935e90e80fe3507589c75eabd5d1635af5fc842f3213","first_computed_at":"2026-05-17T23:39:12.280511Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:39:12.280511Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"vRytmZithn4A0Tv1df48mXvD23g7fGxb7pRKKx0UhgmmHie32rfqUTdLd9GLqfaDqz5P1qLI6p46fmcKBYa9CA==","signature_status":"signed_v1","signed_at":"2026-05-17T23:39:12.281132Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.14084","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:f8525f18b5db78224940a564491b33867f3eba4734a0e385597055f9f1343472","sha256:68d006d379838f136152a67665dcdfa6ecefb2aecf1e23648e698a480429b540"],"state_sha256":"8b5d7c4ba7dd7495647cbdc71b0cbae64fccca0ce0fea14adde4c7aba8422adb"}