pith:QMCAWIL5
Nexus: Same Pretraining Loss, Better Downstream Generalization via Common Minima
Converging to common minima across data sources during pretraining improves downstream generalization even at identical loss values.
arxiv:2604.09258 v2 · 2026-04-10 · cs.LG
Add to your LaTeX paper
\usepackage{pith}
\pithnumber{QMCAWIL5AXFP4Y6GTLMEHQUJT2}
Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge
Record completeness
Claims
Nexus significantly boosts downstream performance, despite achieving the same pretraining loss. Notably, on the 3B model, Nexus reduces the out-of-distribution loss by 0.012 and yields up to a 15.0% accuracy improvement on complex reasoning tasks (e.g., GSM8k).
The geometric closeness of task-specific minima is intrinsically linked to downstream generalization, and that maximizing gradient similarity during optimization produces this closeness.
Nexus optimizer improves LLM downstream performance by converging to common minima across data sources despite identical pretraining loss.
Formal links
Cited by
Receipt and verification
| First computed | 2026-05-28T01:04:39.854864Z |
|---|---|
| Builder | pith-number-builder-2026-05-17-v1 |
| Signature | Pith Ed25519
(pith-v1-2026-05) · public key |
| Schema | pith-number/v1.0 |
Canonical hash
83040b217d05cafe63c69ad843c2899e9fff1b2b8c52e61e1b0ba5ea911e6b9a
Aliases
· · · · ·Agent API
Verify this Pith Number yourself
curl -sH 'Accept: application/ld+json' https://pith.science/pith/QMCAWIL5AXFP4Y6GTLMEHQUJT2 \
| 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: 83040b217d05cafe63c69ad843c2899e9fff1b2b8c52e61e1b0ba5ea911e6b9a
Canonical record JSON
{
"metadata": {
"abstract_canon_sha256": "84a5948012050bc97e76c01cda4ff86e78a223497cc5590b26437950090b5b33",
"cross_cats_sorted": [],
"license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
"primary_cat": "cs.LG",
"submitted_at": "2026-04-10T12:17:18Z",
"title_canon_sha256": "f59b06e55ceb072b5bc53af46f96114d78f57fc9ea1d72a33a834856408e498c"
},
"schema_version": "1.0",
"source": {
"id": "2604.09258",
"kind": "arxiv",
"version": 2
}
}