pith:TFJX65WE
Resting Neurons, Active Insights: Robustifying Activation Sparsity in LLMs via Spontaneity
Spontaneous neurons restore accuracy in activation-sparse large language models by anchoring hidden states to the dense model's distribution.
arxiv:2512.12744 v4 · 2025-12-14 · cs.LG
Add to your LaTeX paper
\usepackage{pith}
\pithnumber{TFJX65WEILIMHMJLN7ZGHBRONJ}
Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge
Record completeness
Claims
We address this issue by reframing activation sparsity as a representational alignment problem and introducing Spontaneous Neurons (SPON), a lightweight mechanism... SPON consistently restores performance, stabilizes latent representations, and preserves generalization.
That a small set of input-independent learnable vectors trained only via distribution matching to the dense model will reliably counteract the distribution shifts induced by activation sparsity without introducing new instabilities or harming generalization on downstream tasks.
SPON adds learnable persistent activation anchors trained via distribution matching to restore LLM accuracy under high activation sparsity by preventing representational distribution shifts.
Formal links
Receipt and verification
| First computed | 2026-05-22T01:03:52.236672Z |
|---|---|
| Builder | pith-number-builder-2026-05-17-v1 |
| Signature | Pith Ed25519
(pith-v1-2026-05) · public key |
| Schema | pith-number/v1.0 |
Canonical hash
99537f76c442d0c3b12b6ff263862e6a6f5452d15ed86b1107df035c64a92d18
Aliases
· · · · ·Agent API
Verify this Pith Number yourself
curl -sH 'Accept: application/ld+json' https://pith.science/pith/TFJX65WEILIMHMJLN7ZGHBRONJ \
| 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: 99537f76c442d0c3b12b6ff263862e6a6f5452d15ed86b1107df035c64a92d18
Canonical record JSON
{
"metadata": {
"abstract_canon_sha256": "d8dec47c6276a44f7b1f062884957537b198d25b37af4e65f5bc3c7cc06ad776",
"cross_cats_sorted": [],
"license": "http://creativecommons.org/licenses/by/4.0/",
"primary_cat": "cs.LG",
"submitted_at": "2025-12-14T15:47:40Z",
"title_canon_sha256": "8f95beb995634cbf53f41f19afbac3831510c7245126bf97add8c16fba9bd337"
},
"schema_version": "1.0",
"source": {
"id": "2512.12744",
"kind": "arxiv",
"version": 4
}
}