Pith Number

pith:KBHRXWCD

pith:2026:KBHRXWCDGU6TL2YHPRJZY4N7YK

not attested not anchored not stored refs resolved

Lifelong Learning in Vision-Language Models: Enhanced EWC with Cross-Modal Knowledge Retention

Hamza Ahmed Durrani, Rafay Suleman Durrani

An enhanced elastic weight consolidation method allows vision-language models to learn tasks sequentially while cutting forgetting rates by 78 percent and keeping image-text alignment intact.

arxiv:2605.12789 v1 · 2026-05-12 · cs.RO

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

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

The framework achieves a 78% reduction in forgetting rates relative to naive sequential training approaches through extensive evaluation testing. The framework also preserves alignment between modalities during sequential learning with only 15% additional computational cost.

C2weakest assumption

That the multi-modal Fisher Information Matrix calculation and adaptive regularization across visual and textual encoders will reliably capture cross-modal dependencies without introducing new forgetting modes or requiring extensive per-task hyperparameter search not described in the abstract.

C3one line summary

Enhanced EWC for LVLMs cuts forgetting rates by 78% versus naive training and keeps visual-textual alignment with 15% extra compute.

References

18 extracted · 18 resolved · 3 Pith anchors

[1] W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., 2021

[2] B., Donahue, J., Luc, P., Miech, A., Barr, I., Hasson, Y ., 2022

[3] Kirkpatrick, J., Pascanu, R., Rabinowitz, N., Veness, J., Desjardins, G., Rusu, A. A., . . . & Hadsell, R. (2017). Overcoming catastrophic forgetting in neural networks.Proceedings of the National Aca 2017

[4] Li, Z., & Hoiem, D. (2017). Learning without forgetting. IEEE Transactions on Pattern Analysis and Machine Intelligence, 40(12), 2935–2947 2017

[5] A., Kolesnikov, A., Sperl, G., & Lampert, C 2017

Receipt and verification

First computed	2026-05-18T03:09:12.958809Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

504f1bd843353d35eb077c539c71bfc289deb605911d36265ed4357e7ce8a0ef

Aliases

arxiv: 2605.12789 · arxiv_version: 2605.12789v1 · doi: 10.48550/arxiv.2605.12789 · pith_short_12: KBHRXWCDGU6T · pith_short_16: KBHRXWCDGU6TL2YH · pith_short_8: KBHRXWCD

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/KBHRXWCDGU6TL2YHPRJZY4N7YK \
  | 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: 504f1bd843353d35eb077c539c71bfc289deb605911d36265ed4357e7ce8a0ef

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "b7fde5828c1be335b2ea5a75aec45329e9fce1bfc77000b0b53f3aef996c4d5a",
    "cross_cats_sorted": [],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.RO",
    "submitted_at": "2026-05-12T22:05:30Z",
    "title_canon_sha256": "185a7c6aebdf31365bab5146f7b132d85df5115f31ca71ee2b6aa8e45587d395"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.12789",
    "kind": "arxiv",
    "version": 1
  }
}