Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Get Started Free →

Cross Platform Verify · yangge521 | DecimalAI

ai mlby yangge521

Cross Platform Verify

yangge521/cross-platform-verify

Source yangge521/dev-skills-zh ↗License MIT

跨平台模型权重验证，确保 YOLO 模型在 NVIDIA T4 和摩尔线程 S4000 间双向可用

SkillScore77Maintained

Automatic

AI Engineer pack

Last verified 8/3/2026

What it changes

Passed 16 of 21 test cases with the skill, and 10 of 21 without it.

Without skill45%

With skill77%

Checks it passed that the model alone missed

Instructs checking image size (imgsz) settings.
Specifies requiring IoU to be greater than 0.99.
Recommends model.save("model.pt") for saving weights.

and 10 more in the full report

7 fixed10 passed either way5 still failing0 broke

Pass-rate lift: +32 pp
Tokens: +15%
Agent turns: 0%
Response time: -18%

Model gemini-3.6-flashMeasured 8/3/2026v117 of 22 cases passed with skill · 21 comparable

Tokens, turns and time describe resource use, not a dollar price.

Use this skill

Save it into your project's skills folder. Claude Code picks it up in its next session.

curl -fsSL --create-dirs -o .claude/skills/yangge521-cross-platform-verify/SKILL.md https://app.decimal.ai/s/yangge521-cross-platform-verify/SKILL.md

Want it pinned and updatable? pip install "decimalai>=0.15", then decimalai skills add yangge521-cross-platform-verify, and start Claude Code with claude --plugin-dir "$(git rev-parse --show-toplevel)/.agents"

1 · One-time — install the CLI (docs ↗)

pip install "decimalai>=0.15"

2 · Get the skill — writes .agents/skills/yangge521-cross-platform-verify/ and pins it in .decimal/skills.lock

decimalai skills add yangge521-cross-platform-verify

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/yangge521-cross-platform-verify/SKILL.md https://app.decimal.ai/s/yangge521-cross-platform-verify/SKILL.md

Raw file: /s/yangge521-cross-platform-verify/SKILL.md ↗ · machine-readable summary at /s/yangge521-cross-platform-verify.json ↗

Safety checks: 3 of 3 passed

Code scan CleanIntent PassedContent Clean

See it in action

One of 7 test cases the skill fixed. Same model, same prompt, with and without the skill.

1 of 7

Prompt

当跨卡对比推理输出时发现重叠度 IoU 稍微低于预期的判定阈值，在检查清单中建议针对图像输入端优先核对哪两项设置？

What was checked, without and with the skill
Checked	Without skill	With skill
Instructs checking image size (imgsz) settings.	Missed	Passed
Instructs checking normalization parameters in input preprocessing.	Missed	Passed

case-18 · gemini-3.6-flash · 8/3/2026 · answers are capped and redacted

What's inside

The instructions your model reads when the skill loads.

SKILL.md81 linessingle file

跨平台模型权重验证

概述

验证 YOLO 模型权重在 NVIDIA T4 和摩尔线程 S4000 之间的双向兼容性，确保在一张卡上训练的模型可以直接在另一张卡上加载推理。

使用场景

训练推理分离：T4 训练，S4000 推理（或反之）
多平台部署：同一模型部署到不同 GPU 集群
GPU 迁移：从 NVIDIA 迁移到国产 GPU
CI/CD 集成：模型发布前自动跨平台验证
供应商评估：验证国产 GPU 对现有模型的兼容性
A/B 对比：同模型不同 GPU 推理结果对比

真实测试数据

四场景验证结果

场景	训练平台	推理平台	权重兼容	结果一致性
1	T4	T4	是	一致
2	T4	S4000	是	一致
3	S4000	T4	是	一致
4	S4000	S4000	是	一致

四场景权重均可用，检测结果完全一致

操作指南

1. 权重保存

训练完成后 model.save("model.pt") 保存。Ultralytics 的 .pt 文件包含模型结构、权重和训练配置，与 GPU 平台无关。

2. 权重加载

在目标平台 YOLO("model.pt") 加载。PyTorch 权重格式跨平台兼容，T4 保存的权重可直接在 S4000 加载。

3. 检测结果一致性验证

Track record

Benchmark

16 / 21

test cases passed, vs 10 / 21 without

Adoption

Users: 0
Times used: 0

Safety and source details

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/1/2026

View 15 scanner checks

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

yangge521 / dev-skills-zh

skills/ai-engineering/cross-platform-verify/SKILL.md

View on GitHub ↗

License MITImported 8/1/2026Version v1

More skills from the same source and nearby in the registry, with their headline benchmark lift.