pith:AOA5CRYT
SinkTrack: Attention Sink based Context Anchoring for Large Language Models
By injecting contextual features into the attention sink token, SinkTrack keeps large language models focused on the original input throughout generation, reducing hallucination and forgetting.
arxiv:2604.10027 v2 · 2026-04-11 · cs.CV
Add to your LaTeX paper
\usepackage{pith}
\pithnumber{AOA5CRYTGYAWL3TOLVLRUFY54X}
Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge
Record completeness
Claims
Experiments demonstrate that SinkTrack mitigates hallucination and context forgetting across both textual (e.g., +21.6% on SQuAD2.0 with Llama3.1-8B-Instruct) and multi-modal (e.g., +22.8% on M3CoT with Qwen2.5-VL-7B-Instruct) tasks.
That consistently high attention to the BOS token can be reliably turned into an effective context anchor simply by injecting input-derived features into its representation, without side effects or the need for per-model tuning.
SinkTrack uses attention sink at the BOS token to anchor LLMs to initial context, reducing hallucination and forgetting with reported gains on benchmarks like SQuAD2.0 and M3CoT.
Receipt and verification
| First computed | 2026-05-20T00:03:10.951599Z |
|---|---|
| Builder | pith-number-builder-2026-05-17-v1 |
| Signature | Pith Ed25519
(pith-v1-2026-05) · public key |
| Schema | pith-number/v1.0 |
Canonical hash
0381d14713360165ee6e5d571a171de5c55bb14cedd8d0a2dd427ec5971eaa16
Aliases
· · · · ·Agent API
Verify this Pith Number yourself
curl -sH 'Accept: application/ld+json' https://pith.science/pith/AOA5CRYTGYAWL3TOLVLRUFY54X \
| 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: 0381d14713360165ee6e5d571a171de5c55bb14cedd8d0a2dd427ec5971eaa16
Canonical record JSON
{
"metadata": {
"abstract_canon_sha256": "9a6b7ffd67c7940fc0dbfaa9ae4eca0cba0eb505ff3e43d86d7c1c2f52926df2",
"cross_cats_sorted": [],
"license": "http://creativecommons.org/licenses/by/4.0/",
"primary_cat": "cs.CV",
"submitted_at": "2026-04-11T04:49:11Z",
"title_canon_sha256": "ba402a401de0d1f987d99155a85e176303d86d3f7eea795b622b9d17c2b8a8ee"
},
"schema_version": "1.0",
"source": {
"id": "2604.10027",
"kind": "arxiv",
"version": 2
}
}