Pith Number

pith:FQXNA77O

pith:2026:FQXNA77ORWYDEG4KRNLSNDLPM7

not attested not anchored not stored refs resolved

NGM: A Plug-and-Play Training-Free Memory Module for LLMs

Caifeng Shan, Chenyang Si, Wenhui Dong, Yuwen Qu

Averaging pretrained token embeddings creates n-gram representations that a cosine-gated injector adds to LLMs without training or extra parameters.

arxiv:2605.16893 v1 · 2026-05-16 · cs.AI

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

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{FQXNA77ORWYDEG4KRNLSNDLPM7}

Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge

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

NGM improves average performance by 0.5 to 1.2 points, with particularly clear gains on code generation and knowledge-intensive tasks (e.g., +3.0 on LiveCodeBench and +3.03 on GPQA for Qwen3-14B).

C2weakest assumption

That directly averaging pretrained token embeddings produces useful n-gram representations that the cosine-gated injector can meaningfully modulate without introducing noise or requiring any learned parameters or additional training.

C3one line summary

NGM is a plug-and-play n-gram memory module that encodes n-grams from pretrained embeddings and gates their injection to improve LLM performance by 0.5-1.2 points on average across eight benchmarks.

References

44 extracted · 44 resolved · 13 Pith anchors

[1] Qwen3-VL Technical Report 2025 · arXiv:2511.21631

[2] Enriching word vec- tors with subword information 2017

[3] Improving language models by retrieving from trillions of tokens 2022

[4] Large language models in machine translation 2007

[5] Are we on the right way for evaluating large vision- language models? Advances in Neural Information Processing Systems , 37:27056–27087, 2024 2024

Formal links

2 machine-checked theorem links

Receipt and verification

First computed	2026-05-20T00:03:28.715557Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

2c2ed07fee8db0321b8a8b57268d6f67eb7357c81972eaca89e7737a1790fdf9

Aliases

arxiv: 2605.16893 · arxiv_version: 2605.16893v1 · doi: 10.48550/arxiv.2605.16893 · pith_short_12: FQXNA77ORWYD · pith_short_16: FQXNA77ORWYDEG4K · pith_short_8: FQXNA77O

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/FQXNA77ORWYDEG4KRNLSNDLPM7 \
  | 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: 2c2ed07fee8db0321b8a8b57268d6f67eb7357c81972eaca89e7737a1790fdf9

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "4e867976f7f559626b33bccc4a190943e4772af7e96f6da5a2c6426838d31d67",
    "cross_cats_sorted": [],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.AI",
    "submitted_at": "2026-05-16T09:12:52Z",
    "title_canon_sha256": "fa3dfc81e90801d2508f65fa03902b664d983b73726720d46dd5570ad6ea5c86"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.16893",
    "kind": "arxiv",
    "version": 1
  }
}