基于 SOC 职业分类
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/Ntizar/MasterMind --skill timesfm-forecast命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
正在显示 SKILL.md
| name | timesfm-forecast |
| description | Usa a pronosticar series temporales con TimesFM 3.0. |
| version | 2.0.0 |
| tags | ["timesfm","forecast","series-temporales","timesfm3","time-series","python"] |
| related_skills | ["timesfm-forecast","qlib-quant","monte-carlo-stock-simulator"] |
⚠️ Corrección 2026-09-05 (auditoría): la v1 usaba
timesfm.TimesFm/model.forecast()(TimesFM v1/v2). La versión actual 3.0 usafrom timesfm3 import TimesFM3Evaluator, ModelConfigyforecaster.predict_batch(...). Install real:pip install timesfm[torch].
Repo: https://github.com/google-research/timesfm (Python, ~31K⭐).
pip install "timesfm[torch]"
from timesfm3 import TimesFM3Evaluator, ModelConfig
model = TimesFM3Evaluator(config=ModelConfig(...))
forecasts = model.predict_batch(inputs, horizon=..., return_quantiles=True)
pip install "timesfm[torch]", no pip install timesfm.TimesFM3Evaluator/predict_batch; no TimesFm/forecast/forecast_with_quantiles (v1/v2).predict_batch(inputs, horizon=N) y comprobar el forecast + quantiles.Usa al medir reps o cadencia desde vídeo con pose.
Usa al predecir relaciones entre objetos en tiempo real.
Patrón de cálculo de sombras solares con Web Workers + Comlink para no bloquear la UI