Pith Number

pith:ECALI4OF

pith:2026:ECALI4OFANEXYE2OSNPJB2AP4N

not attested not anchored not stored refs resolved

CRANE: Constrained Reasoning Injection for Code Agents via Nullspace Editing

Michele Merler, Mingzhi Zhu, Raju Pavuluri, Stacy Patterson

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.

arxiv:2605.14084 v1 · 2026-05-13 · cs.SE · cs.AI · cs.CL

Open paper page JSON Open Graph Bundle Merged state Verified badge What is a Pith Number?

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{ECALI4OFANEXYE2OSNPJB2AP4N}

Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge

Record completeness

1 Bitcoin timestamp

2 Internet Archive

3 Author claim open · sign in to claim

4 Citations open

5 Replications open

✓ Portable graph bundle live · download bundle · merged state

The bundle contains the canonical record plus signed events. A mirror can host it anywhere and recompute the same current state with the deterministic merge algorithm.

Claims

C1strongest 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.

C2weakest 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.

C3one line summary

CRANE merges Instruct and Thinking model checkpoints via constrained nullspace editing to improve code agent reasoning and benchmark performance without retraining.

References

21 extracted · 21 resolved · 7 Pith anchors

[1] Qwen3-Coder-Next Technical Report 2026 · arXiv:2603.00729

[2] DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning · arXiv:2501.12948

[3] Lorahub: Efficient cross-task generalization via dynamic lora composition

[4] SWE-bench: Can Language Models Resolve Real-World GitHub Issues? · arXiv:2310.06770

[5] arXiv preprint arXiv:2505.22113 , year=

Receipt and verification

First computed	2026-05-17T23:39:12.280511Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

2080b471c503497c134e935e90e80fe3507589c75eabd5d1635af5fc842f3213

Aliases

arxiv: 2605.14084 · arxiv_version: 2605.14084v1 · doi: 10.48550/arxiv.2605.14084 · pith_short_12: ECALI4OFANEX · pith_short_16: ECALI4OFANEXYE2O · pith_short_8: ECALI4OF

Agent API

Resolver JSON Graph JSON Events JSON Schema Signing key

Verify this Pith Number yourself

curl -sH 'Accept: application/ld+json' https://pith.science/pith/ECALI4OFANEXYE2OSNPJB2AP4N \
  | jq -c '.canonical_record' \
  | python3 -c "import sys,json,hashlib; b=json.dumps(json.loads(sys.stdin.read()), sort_keys=True, separators=(',',':'), ensure_ascii=False).encode(); print(hashlib.sha256(b).hexdigest())"
# expect: 2080b471c503497c134e935e90e80fe3507589c75eabd5d1635af5fc842f3213

Canonical record JSON

{
  "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
  }
}