Pith Number

pith:NJVNSFNV

pith:2026:NJVNSFNVVR5OYWYFXIQC772ESY

not attested not anchored not stored refs pending

Rewarding the Scientific Process: Process-Level Reward Modeling for Agentic Data Analysis

Huajun Chen, Kewei Xu, Lun Du, Ningyu Zhang, Shuofei Qiao, Yuqi Zhu, Zhisong Qiu

DataPRM improves AI data analysis agents by actively probing execution states to catch silent errors and applying ternary rewards that separate fixable mistakes from fatal ones.

arxiv:2604.24198 v2 · 2026-04-27 · cs.CL · cs.AI · cs.CE · cs.LG · cs.MA

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

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{NJVNSFNVVR5OYWYFXIQC772ESY}

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

DataPRM improves downstream policy LLMs by 7.21% on ScienceAgentBench and 11.28% on DABStep using Best-of-N inference; integrating DataPRM into Reinforcement Learning yields 78.73% on DABench and 64.84% on TableBench.

C2weakest assumption

That the performance gains are attributable to the environment-aware probing and reflection-aware ternary strategy rather than to the quality or diversity of the 8K training instances or to unstated differences in baselines and evaluation protocols.

C3one line summary

DataPRM is a new process reward model for data analysis agents that detects silent errors via environment interaction and ternary rewards, yielding 7-11% gains on benchmarks and further RL improvements.

Cited by

1 paper in Pith

From Table to Cell: Attention for Better Reasoning with TABALIGN

Receipt and verification

First computed	2026-06-23T02:12:49.414552Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

6a6ad915b5ac7aec5b05ba202fff4496298529e81f85baff66f9f27cace4cfb6

Aliases

arxiv: 2604.24198 · arxiv_version: 2604.24198v2 · doi: 10.48550/arxiv.2604.24198 · pith_short_12: NJVNSFNVVR5O · pith_short_16: NJVNSFNVVR5OYWYF · pith_short_8: NJVNSFNV

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/NJVNSFNVVR5OYWYFXIQC772ESY \
  | 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: 6a6ad915b5ac7aec5b05ba202fff4496298529e81f85baff66f9f27cace4cfb6

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "ce10f76542bdf90fa13fd1e77811802f36b8ad814a1003589a35a368a42c41ae",
    "cross_cats_sorted": [
      "cs.AI",
      "cs.CE",
      "cs.LG",
      "cs.MA"
    ],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.CL",
    "submitted_at": "2026-04-27T09:00:30Z",
    "title_canon_sha256": "45172b74396e67fd503eb7c2e5237f548e9d36c5d218da62f41e187452af1002"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2604.24198",
    "kind": "arxiv",
    "version": 2
  }
}