Pith Number

pith:2BHG65M3

pith:2026:2BHG65M3XUU5MWQG2ZNNN23SCP

not attested not anchored not stored refs pending

Learning Evidence Highlighting for Frozen LLMs

Chonglin Sun, Fei Tian, Frank Shyu, Jian Li, Luke Simon, Mingfu Liang, Sandeep Pandey, Shaoang Li, Xiaohan Wei, Xi Liu, Yanhang Shi, Yufei Li, Yunchen Pu

A lightweight actor learns to insert highlight tags around key evidence to improve frozen LLMs on long contexts.

arxiv:2604.22565 v2 · 2026-04-24 · cs.CL · cs.AI

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

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{2BHG65M3XUU5MWQG2ZNNN23SCP}

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

Across sequential recommendation and long-context question answering, HiLight consistently improves performance over strong prompt-based and automated prompt-optimization baselines. The learned emphasis policy transfers zero-shot to both smaller and larger unseen Solver families, including an API-based Solver.

C2weakest assumption

That inserting minimal highlight tags around pivotal spans guides the frozen Solver's reasoning toward better evidence use without introducing new biases or distortions, and that the RL policy learns reusable evidence structure rather than overfitting to the training tasks or solvers.

C3one line summary

HiLight uses reinforcement learning to train a lightweight actor that inserts highlight tags around pivotal evidence for frozen LLMs, improving sequential recommendation and long-context QA while transferring zero-shot to unseen models.

Receipt and verification

First computed	2026-06-10T01:08:35.961254Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

d04e6f759bbd29d65a06d65ad6eb7213e321fe20da05a1b596ba08a621a5ad4f

Aliases

arxiv: 2604.22565 · arxiv_version: 2604.22565v2 · doi: 10.48550/arxiv.2604.22565 · pith_short_12: 2BHG65M3XUU5 · pith_short_16: 2BHG65M3XUU5MWQG · pith_short_8: 2BHG65M3

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/2BHG65M3XUU5MWQG2ZNNN23SCP \
  | 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: d04e6f759bbd29d65a06d65ad6eb7213e321fe20da05a1b596ba08a621a5ad4f

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "e30b9a367574411f880bda2d02d888fefad2b0f878c7552e62fe0c7a12d5f7fb",
    "cross_cats_sorted": [
      "cs.AI"
    ],
    "license": "http://creativecommons.org/licenses/by/4.0/",
    "primary_cat": "cs.CL",
    "submitted_at": "2026-04-24T13:57:19Z",
    "title_canon_sha256": "7ed872c718888caa06361ac066246300491044843a2229199671311cea33eef4"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2604.22565",
    "kind": "arxiv",
    "version": 2
  }
}