Pith Number

pith:WT2B2WJP

pith:2025:WT2B2WJP3OXQYVOSO2KO6W37AP

not attested not anchored not stored refs resolved

Enhancing LLM-Based Neural Network Generation: Few-Shot Prompting and Efficient Validation for Automated Architecture Design

Avi Goyal, Chandini Vysyaraju, Dmitry Ignatov, Radu Timofte, Raghuvir Duvvuri

Three examples in few-shot prompts let LLMs generate the most balanced neural architectures for vision tasks while a simple hash check speeds validation by 100 times.

arxiv:2512.24120 v2 · 2025-12-30 · cs.CV · 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{WT2B2WJP3OXQYVOSO2KO6W37AP}

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

Using n = 3 examples best balances architectural diversity and context focus for vision tasks. Whitespace-Normalized Hash Validation provides a 100x speedup over AST parsing and prevents redundant training of duplicate architectures.

C2weakest assumption

That the observed performance differences across prompting regimes are driven by the number of examples rather than by uncontrolled variation in LLM sampling, dataset splits, or training hyperparameters.

C3one line summary

Three-example few-shot prompting optimizes LLM-generated vision architectures while a whitespace-normalized hash provides 100x faster duplicate detection than AST parsing across seven benchmarks.

References

30 extracted · 30 resolved · 8 Pith anchors

[1] AUGMENTGEST: Can random data cropping augmentation boost gesture recognition performance?arXiv preprint 2025

[2] Brown, Benjamin Mann, Nick Ryder, Melanie Sub- biah, Jared Kaplan, Prafulla Dhariwal, et al 1901

[3] Evaluating Large Language Models Trained on Code 2021 · arXiv:2107.03374

[4] Program Synthesis with Large Language Models 2022 · arXiv:2108.07732

[5] DeepSeek-Coder: When the Large Language Model Meets Programming -- The Rise of Code Intelligence 2024 · arXiv:2401.14196

Cited by

2 papers in Pith

Closed-Loop LLM Discovery of Non-Standard Channel Priors in Vision Models

Delta-Based Neural Architecture Search: LLM Fine-Tuning via Code Diffs

Receipt and verification

First computed	2026-06-02T02:04:12.786101Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

b4f41d592fdbaf0c55d27694ef5b7f03f03e6f91c60d07bb18897263dc298645

Aliases

arxiv: 2512.24120 · arxiv_version: 2512.24120v2 · doi: 10.48550/arxiv.2512.24120 · pith_short_12: WT2B2WJP3OXQ · pith_short_16: WT2B2WJP3OXQYVOS · pith_short_8: WT2B2WJP

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/WT2B2WJP3OXQYVOSO2KO6W37AP \
  | 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: b4f41d592fdbaf0c55d27694ef5b7f03f03e6f91c60d07bb18897263dc298645

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "035e2a7307fd7baaf807caed199e57d0ff222816b1e5b2d57d969f5c02244eee",
    "cross_cats_sorted": [
      "cs.AI"
    ],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.CV",
    "submitted_at": "2025-12-30T10:01:55Z",
    "title_canon_sha256": "cef0f864368c9e74340f02a0e0288eeb6143455ece25681089161a2f1d5d8acc"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2512.24120",
    "kind": "arxiv",
    "version": 2
  }
}