pith:XARKISWT
CompactAttention: Accelerating Chunked Prefill with Block-Union KV Selection
Block-union KV selection builds minimal tables so chunked prefill attention runs up to 2.72 times faster while staying close to dense accuracy.
arxiv:2605.16839 v1 · 2026-05-16 · cs.CL
Add to your LaTeX paper
\usepackage{pith}
\pithnumber{XARKISWTWL77GOF3REB7SEHZXM}
Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge
Record completeness
Claims
On LLaMA-3.1-8B-Instruct, CompactAttention maintains accuracy close to dense attention on the RULER benchmark while delivering up to 2.72× attention speedup at 128K context length under chunked prefill.
The construction of GQA-aware per-group KV block tables via Q-block union and intra-group union produces the minimal tables that preserve all KV blocks selected by the input masks under paged execution constraints without missing query-specific entries.
CompactAttention accelerates chunked-prefill attention via Block-Union KV Selection, delivering up to 2.72x speedup at 128K context on LLaMA-3.1-8B while matching dense accuracy on RULER.
References
Formal links
Receipt and verification
| First computed | 2026-05-20T00:03:25.502780Z |
|---|---|
| Builder | pith-number-builder-2026-05-17-v1 |
| Signature | Pith Ed25519
(pith-v1-2026-05) · public key |
| Schema | pith-number/v1.0 |
Canonical hash
b822a44ad3b2fff338bb8903f910f9bb14f361a79e1e056d6778f43b63fabd8f
Aliases
· · · · ·Agent API
Verify this Pith Number yourself
curl -sH 'Accept: application/ld+json' https://pith.science/pith/XARKISWTWL77GOF3REB7SEHZXM \
| 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: b822a44ad3b2fff338bb8903f910f9bb14f361a79e1e056d6778f43b63fabd8f
Canonical record JSON
{
"metadata": {
"abstract_canon_sha256": "7ae9a3ee0b125d2213d63e83abef7aacf5a01522096f9c596bd9658a473937a5",
"cross_cats_sorted": [],
"license": "http://creativecommons.org/licenses/by/4.0/",
"primary_cat": "cs.CL",
"submitted_at": "2026-05-16T06:47:41Z",
"title_canon_sha256": "737bfbe36fe943f6268813677dd619f9410ed42f63c6c0a7ea40ef1f71ee028d"
},
"schema_version": "1.0",
"source": {
"id": "2605.16839",
"kind": "arxiv",
"version": 1
}
}