Pith Number

pith:D2PMXLZT

pith:2026:D2PMXLZTVS5YAYHHJS6RTGU3RJ

not attested not anchored not stored refs resolved

Evaluation Drift in LLM Personality Induction: Are We Moving the Goalpost?

Iyiola E. Olatunji, Jacques Klein, Prateek Rajput, Tegawend\'e F. Bissyand\'e, Yewei Song

Fine-tuning stabilizes LLM personality questionnaire scores but full-profile accuracy stays near chance.

arxiv:2605.16996 v1 · 2026-05-16 · cs.CL

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

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{D2PMXLZTVS5YAYHHJS6RTGU3RJ}

Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more

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

Our results demonstrate that fine-tuning consistently reduces variance in questionnaire responses across five models, directly mitigating the evaluation fragility reported in pre-trained models. However, this newfound stability reveals a more fundamental limitation: accuracy on the full five-dimensional profile remains near chance, even when single-trait scores improve.

C2weakest assumption

The IPIP-NEO questionnaire responses from LLMs validly and comprehensively measure the induced personality profile, and that the unguided essays contain sufficient trait-relevant cues to support faithful induction of the target five-dimensional profile.

C3one line summary

Fine-tuning LLMs on essays reduces variance in IPIP-NEO responses across models but does not raise full five-trait profile accuracy above near-chance levels from unguided text.

References

16 extracted · 16 resolved · 3 Pith anchors

[1] InFindings of the Associa- tion for Computational Linguistics: EMNLP 2023, pages 2370–2386, Singapore 2023

[2] Hans Christian, Derwin Suhartono, Andry Chowanda, and Kamal Z Zamli

[3] The Llama 3 Herd of Models · arXiv:2407.21783

[4] Matej Gjurković and Jan Šnajder

[5] In2011 IEEE third international conference on privacy, security, risk and trust and 2011 IEEE third international conference on social computing, pages 149–156 2011

Formal links

2 machine-checked theorem links

Receipt and verification

First computed	2026-05-20T00:03:35.087993Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

1e9ecbaf33acbb8060e74cbd199a9b8a7f929a4081f17701b25eb436ba9fccf7

Aliases

arxiv: 2605.16996 · arxiv_version: 2605.16996v1 · doi: 10.48550/arxiv.2605.16996 · pith_short_12: D2PMXLZTVS5Y · pith_short_16: D2PMXLZTVS5YAYHH · pith_short_8: D2PMXLZT

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/D2PMXLZTVS5YAYHHJS6RTGU3RJ \
  | 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: 1e9ecbaf33acbb8060e74cbd199a9b8a7f929a4081f17701b25eb436ba9fccf7

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "90f15a7fbc944fdfa408cbe6b145e55b409b078e064ac2408734a63f14a81414",
    "cross_cats_sorted": [],
    "license": "http://creativecommons.org/licenses/by/4.0/",
    "primary_cat": "cs.CL",
    "submitted_at": "2026-05-16T13:44:06Z",
    "title_canon_sha256": "41653e4a97fc34708a64fc4f9d2b0c4c6a70f0b64633c425c1dbb7189f8d59ad"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.16996",
    "kind": "arxiv",
    "version": 1
  }
}