Pith Number

pith:UXVIOBJL

pith:2026:UXVIOBJL7BGXGDYQ4OV7K4ZQ7X

not attested not anchored not stored refs pending

Model-Dowser: Data-Free Importance Probing to Mitigate Catastrophic Forgetting in Multimodal Large Language Models

Daeyoung Kim, Hyeontaek Hwang, Nguyen Dinh Son

Model-Dowser uses a joint importance score from weights, activations and sensitivities to freeze key parameters during fine-tuning and thereby reduce catastrophic forgetting in multimodal large language models.

arxiv:2602.04509 v6 · 2026-02-04 · cs.CL

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

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{UXVIOBJL7BGXGDYQ4OV7K4ZQ7X}

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

Comprehensive experiments on two representative MLLMs, LLaVA and NVILA, demonstrate that Model-Dowser effectively mitigates catastrophic forgetting and consistently outperforms prior methods, while remaining resource-efficient and scalable to multi-billion-parameter models.

C2weakest assumption

The joint importance score computed from weight magnitudes, input activations, and output sensitivities before any downstream adaptation accurately identifies the parameters whose preservation is necessary and sufficient to maintain pretrained generalization.

C3one line summary

Model-Dowser computes a joint importance score from weights, activations, and sensitivities to selectively update parameters during fine-tuning, reducing catastrophic forgetting in MLLMs like LLaVA and NVILA.

Formal links

2 machine-checked theorem links

Cited by

1 paper in Pith

From Structure to Synergy: A Survey of Vision-Language Perception Paradigm Evolution in Multimodal Large Language Models

Receipt and verification

First computed	2026-05-22T01:03:56.408571Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

a5ea87052bf84d730f10e3abf57330fdce3364d224e82b4f430ecc7ce8684085

Aliases

arxiv: 2602.04509 · arxiv_version: 2602.04509v6 · doi: 10.48550/arxiv.2602.04509 · pith_short_12: UXVIOBJL7BGX · pith_short_16: UXVIOBJL7BGXGDYQ · pith_short_8: UXVIOBJL

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/UXVIOBJL7BGXGDYQ4OV7K4ZQ7X \
  | 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: a5ea87052bf84d730f10e3abf57330fdce3364d224e82b4f430ecc7ce8684085

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "657e86e1bf47656ce7f11f4e6f13f8ab2e180f969c87b94dc907a9bec154d45a",
    "cross_cats_sorted": [],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.CL",
    "submitted_at": "2026-02-04T12:56:27Z",
    "title_canon_sha256": "040e46c73c52115a02906292e459c01311663bfaa2c383bb8d3679b53b610977"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2602.04509",
    "kind": "arxiv",
    "version": 6
  }
}