pith:YIPXLBFK
Controllable and Verifiable Process Data Synthesis for Process Reward Models
A synthesis method builds controllable process supervision data by injecting template-aware errors into symbolic reasoning chains, recomputing trajectories, and translating them to natural language for training process reward models.
arxiv:2605.02395 v2 · 2026-05-04 · cs.AI
Add to your LaTeX paper
\usepackage{pith}
\pithnumber{YIPXLBFKIR64FLKSD77A5MQPXS}
Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge
Record completeness
Claims
Experiments show that the synthesized data improve Best-of-8 reranking on logical reasoning benchmarks and transfer to mathematical reasoning. Step-level evaluation further shows that first-error localization remains substantially more challenging than overall step classification.
The assumption that template-aware errors injected into symbolic chains and then recomputed produce trajectories whose error patterns and consistency properties transfer meaningfully to natural-language reasoning processes used in real PRM training.
A controllable synthesis method creates prefix-invalid yet trajectory-consistent process supervision data for training and evaluating process reward models by injecting verifiable errors into symbolic reasoning chains.
Formal links
Receipt and verification
| First computed | 2026-06-05T01:14:39.903081Z |
|---|---|
| Builder | pith-number-builder-2026-05-17-v1 |
| Signature | Pith Ed25519
(pith-v1-2026-05) · public key |
| Schema | pith-number/v1.0 |
Canonical hash
c21f7584aa447dc2ad521ffe0eb20fbcb742473a92848b82c03b10aea0fb069d
Aliases
· · · · ·Agent API
Verify this Pith Number yourself
curl -sH 'Accept: application/ld+json' https://pith.science/pith/YIPXLBFKIR64FLKSD77A5MQPXS \
| 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: c21f7584aa447dc2ad521ffe0eb20fbcb742473a92848b82c03b10aea0fb069d
Canonical record JSON
{
"metadata": {
"abstract_canon_sha256": "0f4f310104438fccb75c5a90592ba01100485f452c56fb76290f326fd8fed9c9",
"cross_cats_sorted": [],
"license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
"primary_cat": "cs.AI",
"submitted_at": "2026-05-04T09:36:57Z",
"title_canon_sha256": "c149d2cbea3f31883b2110a96f47abfdbf165eee56ece73506d976a1e1b3bc08"
},
"schema_version": "1.0",
"source": {
"id": "2605.02395",
"kind": "arxiv",
"version": 2
}
}