Pith Number

pith:U4ZQRXTA

pith:2026:U4ZQRXTAQTMTDQE4WS66F2FVVP

not attested not anchored not stored refs resolved

Data Agent: Learning to Select Data via End-to-End Dynamic Optimization

Baile Xu, Fangjian Su, Furao Shen, Hai Gan, Jie Li, Soujanya Poria, Suorong Yang, Ziqi Ye

Data Agent learns to select training samples dynamically as a sequential decision problem guided by evolving loss and uncertainty rewards.

arxiv:2603.07433 v2 · 2026-03-08 · cs.LG · cs.CV

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

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{U4ZQRXTAQTMTDQE4WS66F2FVVP}

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

Data Agent consistently accelerates training while preserving or improving performance, e.g., reducing costs by over 50% on ImageNet-1k and MMLU with lossless performance.

C2weakest assumption

That a composite reward combining loss-based difficulty and confidence-based uncertainty, together with a tuning-free adaptive weighting mechanism, can reliably capture the evolving utility of each sample throughout training across diverse tasks and architectures.

C3one line summary

Data Agent learns a co-evolving sample selection policy end-to-end that accelerates training by over 50% on ImageNet-1k and MMLU with no performance loss.

References

19 extracted · 19 resolved · 8 Pith anchors

[1] GPT-4 Technical Report · arXiv:2303.08774

[2] Gradnorm: Gradient normalization for adaptive loss balancing in deep multitask networks · arXiv:1711.02257

[3] An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale 2010 · arXiv:2010.11929

[4] Length-Controlled AlpacaEval: A Simple Way to Debias Automatic Evaluators · arXiv:2404.04475

[5] Measuring Massive Multitask Language Understanding 2009 · arXiv:2009.03300

Formal links

2 machine-checked theorem links

Receipt and verification

First computed	2026-05-18T03:09:22.958437Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

a73308de6084d931c09cb4bde2e8b5abd154fcf377459d97b3887ca245db10d2

Aliases

arxiv: 2603.07433 · arxiv_version: 2603.07433v2 · doi: 10.48550/arxiv.2603.07433 · pith_short_12: U4ZQRXTAQTMT · pith_short_16: U4ZQRXTAQTMTDQE4 · pith_short_8: U4ZQRXTA

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/U4ZQRXTAQTMTDQE4WS66F2FVVP \
  | 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: a73308de6084d931c09cb4bde2e8b5abd154fcf377459d97b3887ca245db10d2

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "1c86de336f20c56f2fb81bfefa1043abefce8647b7745cc9d990d5953f398ca3",
    "cross_cats_sorted": [
      "cs.CV"
    ],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.LG",
    "submitted_at": "2026-03-08T03:10:39Z",
    "title_canon_sha256": "0a17e150362ea75f1b651945cd110cac508b18037eaff452b8329cd41f0fb872"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2603.07433",
    "kind": "arxiv",
    "version": 2
  }
}