Borrowing it
Nothing to install: this file belongs to ayutaz/piper-plus. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/ayutaz/piper-plus/dev/.claude/skills/publish-model/SKILL.mdgit clone --depth 1 https://github.com/ayutaz/piper-plusWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/ayutaz/piper-plus/publish-model)<a href="https://agentmods.dev/skills/ayutaz/piper-plus/publish-model"><img src="https://agentmods.dev/badge/skills/ayutaz/piper-plus/publish-model.svg" alt="Measured on agentmods" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00094 | $0.01615 |
| Opus 5 | $0.00047 | $0.00807 |
| Sonnet 5 | $0.00019 | $0.00323 |
| Haiku 4.5 | $0.00009 | $0.00161 |
Grade A, and why
publish-model scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 7d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 137 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Model Publish Pipeline Skill
Lightning checkpoint から HuggingFace 公開までを 1 つの skill に集約。 現状 4 ステップ手動運用 (export → infer test → RTF bench → HF push) で、 工程間の引き継ぎミスが頻発する。
memory feedback_merge_caution.md に従い、 publish (HuggingFace への push) は確認後に明示実行。 デフォルトは export + sanity + bench までで停止。
引数
$1(必須): checkpoint パス、 例/data/piper/output-tsukuyomi-finetune-6lang-v2/last.ckpt--output PATH: 出力 ONNX パス (デフォルト: checkpoint 隣に.onnx)--repo OWNER/REPO: HuggingFace repo 名 (デフォルト:ayousanz/<voice-key>)--skip-benchmark: フェーズ 4 を skip--apply: フェーズ 5 (HuggingFace upload) も実行
現在の状態
- ブランチ: !
git rev-parse --abbrev-ref HEAD - 引数: $ARGUMENTS
フェーズ 1: 事前検査
# Checkpoint 存在 / サイズ確認
ls -la "$1"
file "$1" # PyTorch Lightning checkpoint であることを確認
# Contract gate (drift 検出して bump 前 fail)
uv run python scripts/check_onnx_export_contract.py 2>&1 | tail -3
uv run python scripts/check_phoneme_set_version.py 2>&1 | tail -3
フェーズ 2: ONNX export
CLAUDE.md の推奨設定 (FP16 + EMA + stochastic + emb_lang 自動統一) を使う:
CUDA_VISIBLE_DEVICES="" uv run python -m piper_train.export_onnx \
"$CKPT_PATH" "$OUTPUT_ONNX"
出力後の verification:
# ONNX checker / shape inference
uv run python -c "import onnx; m=onnx.load('$OUTPUT_ONNX'); onnx.checker.check_model(m); onnx.shape_inference.infer_shapes(m)"
# Size 確認 (FP16 で ~50% 削減されているか)
ls -la "$OUTPUT_ONNX"
フェーズ 3: Inference sanity check
# JSONL 1 行入力で推論テスト
echo '{"phoneme_ids": [1, 2, 3, 4, 5], "speaker_id": 0}' | \
CUDA_VISIBLE_DEVICES="" uv run python -m piper_train.infer_onnx \
--model "$OUTPUT_ONNX" --output-dir /tmp/sanity
# 出力 wav の sanity 確認
ls -la /tmp/sanity/*.wav
無音 / クリップ / 形状不一致を catch。
フェーズ 4: RTF benchmark (optional)
uv run python tools/benchmark/run_benchmark.py \
--model "$OUTPUT_ONNX" --warmup 5 --runs 30 \
--output /tmp/rtf_$(basename "$OUTPUT_ONNX" .onnx).json
baseline (README.md の Benchmark 表、 Xeon E5-2650 v4 / 25 phoneme 英文 / 27ms) と比較し、 ±30% 以内なら OK、 大きい drift があれば warning。
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 7d ago First seen · 137 lines · 94 tokens per session scan A cf726fec46e0
publish-model is a skill published in the GitHub repository ayutaz/piper-plus (204 stars, last pushed 5d ago), licensed MIT. It adds 94 tokens to every session and 1,615 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
physicsnemo-shard-tensor
Official NVIDIA-authored guidance for PhysicsNeMo ShardTensor domain parallelism — integrate domain parallelism into training/inference scripts (new or existing) with DDP or FSDP2, write and register shard patches to enable new layers/ops, and bootstrap multi-GPU correctness tests. Use when working with ShardTensor…
physicsnemo-discover
Official NVIDIA-authored guidance for navigating PhysicsNeMo — pick the model, datapipe, or example for a SciML/AI4Science task (surrogates, forecasting, downscaling, physics-informed, inverse, generative). Points at existing files via live repo search; never writes code. Do NOT use for installation or environment…
single-cell-multi-omics-analysis-scvi
Probabilistic deep learning framework for single-cell multi-omics data analysis. Use this skill when: (1) Analyzing single-cell RNA-seq data with batch correction, (2) Integrating multi-modal data (CITE-seq, ATAC-seq, multi-omics), (3) Performing cell type annotation with scANVI, (4) Spatial transcriptomics…
deep-learning
Use when training or debugging a neural net in PyTorch — the forward/loss/backward/step loop and its silent bugs, mixed precision (AMP), AdamW/LR schedules, DDP/FSDP/ZeRO, checkpoints and seeds. NOT LoRA/QLoRA on a pretrained LLM (that is finetuning), NOT tabular sklearn/XGBoost (that is machine-learning), NOT…
graphsignal-profiler
Set up the Graphsignal Profiler for inference workloads — vLLM, SGLang, PyTorch, and dstack services. Use when the user wants GPU profiling, tracing, or monitoring for inference, asks about graphsignal-run or graphsignal.watch(), or asks about CUPTI / Prometheus / OTLP setup.
atc-model-converter
Complete toolkit for Huawei Ascend NPU model conversion and end-to-end inference adaptation. Workflow 1 auto-discovers input shapes and parameters from user source code. Workflow 2 exports PyTorch models to ONNX. Workflow 3 converts ONNX to .om via ATC with multi-CANN version support. Workflow 4 adapts the user's full…