Pith Number

pith:3ZAK63EJ

pith:2026:3ZAK63EJXKY5D2UZPLCZM3PUMN

not attested not anchored not stored refs resolved

Respecting Self-Uncertainty in On-Policy Self-Distillation for Efficient LLM Reasoning

Conghui He, Junlong Ke, Linfeng Zhang, Weijia Li, Zichen Wen

An entropy confidence gate that down-weights uncertain tokens improves the accuracy-length trade-off in on-policy self-distillation for LLM reasoning.

arxiv:2605.13255 v1 · 2026-05-13 · 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{3ZAK63EJXKY5D2UZPLCZM3PUMN}

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

Experiments with Qwen3-4B and Qwen3-8B in thinking mode show that EGRSD and CL-EGRSD advance the accuracy-length frontier among the compared trainable methods.

C2weakest assumption

That selectively down-weighting high-entropy tokens via the teacher-entropy confidence gate improves net reasoning quality without discarding critical information that only appears in uncertain positions.

C3one line summary

EGRSD and CL-EGRSD advance the accuracy-length frontier in LLM reasoning by entropy-guided weighting of token-level distillation signals from the teacher.

References

27 extracted · 27 resolved · 10 Pith anchors

[1] arXiv preprint arXiv:2505.16400 , year=

[2] Training Verifiers to Solve Math Word Problems · arXiv:2110.14168

[3] OpenThoughts: Data Recipes for Reasoning Models · arXiv:2506.04178

[4] Entropy-aware on-policy distillation of language models

[5] Why Does Self-Distillation (Sometimes) Degrade the Reasoning Capability of LLMs? · arXiv:2603.24472

Cited by

2 papers in Pith

A Brief Overview: On-Policy Self-Distillation In Large Language Models

Receipt and verification

First computed	2026-05-18T02:44:49.396514Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

de40af6c89bab1d1ea997ac5966df46358ad9897dcee5fdb409fc775af3e9699

Aliases

arxiv: 2605.13255 · arxiv_version: 2605.13255v1 · doi: 10.48550/arxiv.2605.13255 · pith_short_12: 3ZAK63EJXKY5 · pith_short_16: 3ZAK63EJXKY5D2UZ · pith_short_8: 3ZAK63EJ

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/3ZAK63EJXKY5D2UZPLCZM3PUMN \
  | 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: de40af6c89bab1d1ea997ac5966df46358ad9897dcee5fdb409fc775af3e9699

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "00baede65b540590aefdf9d8125b22901bb17af42eb49249664b3b492e4e4f1d",
    "cross_cats_sorted": [],
    "license": "http://creativecommons.org/licenses/by-nc-nd/4.0/",
    "primary_cat": "cs.AI",
    "submitted_at": "2026-05-13T09:38:20Z",
    "title_canon_sha256": "3fd4635281c5f14f852c07633698fc5e65cf60e63605afa04e829ceeb7619bef"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.13255",
    "kind": "arxiv",
    "version": 1
  }
}