pith. sign in
Pith Number

pith:RJWGEDU6

pith:2026:RJWGEDU6KMOZXTCTILE4QYFPCA
not attested not anchored not stored refs resolved

Knowledge Beyond Language: Bridging the Gap in Multilingual Machine Unlearning Evaluation

Hyeonjin Kim, Kyomin Hwang, Nojun Kwak, Sangyeon Cho

Two metrics called KSS and KPS measure how consistently unlearning removes information across languages in multilingual LLMs.

arxiv:2605.14404 v1 · 2026-05-14 · cs.CL

Add to your LaTeX paper
\usepackage{pith}
\pithnumber{RJWGEDU6KMOZXTCTILE4QYFPCA}

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

We propose two metrics to evaluate the information spread across languages: the Knowledge Separability Score (KSS) and the Knowledge Persistence Score (KPS). KSS measures the overall unlearning quality across multiple languages, while KPS more specifically aims to assess consistent removal of information among different language pairs.

C2weakest assumption

That KSS and KPS accurately capture cross-linguistic information distribution and validly measure unlearning quality, without requiring external validation against actual leakage rates or human judgments of forgetting.

C3one line summary

New metrics KSS and KPS are introduced to evaluate multilingual machine unlearning quality and cross-language consistency in LLMs, addressing limitations of single-language evaluation protocols.

References

55 extracted · 55 resolved · 12 Pith anchors

[1] Aho and Jeffrey D 1972
[2] Publications Manual , year = "1983", publisher = 1983
[3] Chandra and Dexter C 1981 · doi:10.1145/322234.322243
[4] Scalable training of
[5] Dan Gusfield , title =. 1997 1997

Formal links

2 machine-checked theorem links

Receipt and verification
First computed 2026-05-17T23:39:07.448278Z
Builder pith-number-builder-2026-05-17-v1
Signature Pith Ed25519 (pith-v1-2026-05) · public key
Schema pith-number/v1.0

Canonical hash

8a6c620e9e531d9bcc5342c9c860af10082c7cceb001d22140a292f05789b546

Aliases

arxiv: 2605.14404 · arxiv_version: 2605.14404v1 · doi: 10.48550/arxiv.2605.14404 · pith_short_12: RJWGEDU6KMOZ · pith_short_16: RJWGEDU6KMOZXTCT · pith_short_8: RJWGEDU6
Agent API
Verify this Pith Number yourself
curl -sH 'Accept: application/ld+json' https://pith.science/pith/RJWGEDU6KMOZXTCTILE4QYFPCA \
  | 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: 8a6c620e9e531d9bcc5342c9c860af10082c7cceb001d22140a292f05789b546
Canonical record JSON
{
  "metadata": {
    "abstract_canon_sha256": "565d8c4e9b283ddf40e9c1c72d03e2569bfdf6a7ba8433400dbd8d55c39a297e",
    "cross_cats_sorted": [],
    "license": "http://creativecommons.org/licenses/by/4.0/",
    "primary_cat": "cs.CL",
    "submitted_at": "2026-05-14T05:45:24Z",
    "title_canon_sha256": "a5784fa4ec03b091b35f02bdeff1969c42fc74e6070b7cdcb60aa7364316475a"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.14404",
    "kind": "arxiv",
    "version": 1
  }
}