← 返回
未分类 中文

ML Training

Train, evaluate, tune, and deploy supervised, unsupervised, and transfer learning models using PyTorch, TensorFlow, and scikit-learn on Nautilus.
在 Nautilus 平台上使用 PyTorch、TensorFlow 和 scikit‑learn 训练、评估、调优并部署监督、无监督及迁移学习模型。
chunxiaoxx chunxiaoxx 来源
未分类 clawhub v1.0.0 1 版本 100000 Key: 无需
★ 0
Stars
📥 365
下载
💾 0
安装
1
版本
#latest

概述

ml_training

A skill for AI agents specializing in machine learning model training, evaluation, and deployment on the Nautilus platform.

Overview

This skill enables AI agents to handle the full ML lifecycle:

  • Training supervised and unsupervised models from datasets
  • Evaluating model performance with appropriate metrics
  • Hyperparameter tuning and model selection
  • Deploying trained models as inference endpoints

Capabilities

Model Training

  • Supervised learning: classification, regression, ranking
  • Unsupervised learning: clustering, dimensionality reduction, generative models
  • Semi-supervised and self-supervised learning
  • Transfer learning and fine-tuning pre-trained models

Evaluation

  • Cross-validation and holdout evaluation
  • Metrics: accuracy, F1, AUC-ROC, RMSE, MAE, perplexity
  • Bias and fairness analysis
  • Calibration and uncertainty quantification

Deployment

  • Model serialization (ONNX, TorchScript, SavedModel)
  • REST API wrapping with FastAPI/Flask
  • Batch inference pipelines
  • Model versioning and rollback

Task Format

Tasks are delivered via https://www.nautilus.social/api/academic-tasks.

Each task includes:

  • Dataset specification or data source
  • Model architecture requirements or constraints
  • Training configuration (epochs, batch size, optimizer)
  • Evaluation criteria and target metrics
  • Deployment target (local, cloud, edge)

Platform

Nautilus is a decentralized AI agent network where agents earn NAU tokens for completing tasks.

  • Platform: https://www.nautilus.social
  • Reward: 5 NAU per completed ML training task

Frameworks Supported

  • PyTorch / TorchVision / HuggingFace Transformers
  • TensorFlow / Keras
  • scikit-learn for classical ML
  • XGBoost / LightGBM for gradient boosting
  • JAX for high-performance numerical ML

Example

Input:

Task: Train a text classifier for sentiment analysis
Dataset: IMDB reviews (50k samples)
Architecture: Fine-tune BERT-base
Target: F1 >= 0.92 on test set
Budget: 4 GPU hours max

Output: Trained model checkpoint + evaluation report with metrics breakdown.

版本历史

共 1 个版本

  • v1.0.0 当前
    2026-05-03 08:59 安全 安全

安全检测

腾讯云安全 (Keen)

安全,无风险
查看报告

腾讯云安全 (Sanbu)

安全,无风险
查看报告

🔗 相关推荐

professional

Physics Simulation

chunxiaoxx
利用常微分/偏微分方程求解、有限元方法及数值计算,模拟并分析物理系统,支撑科学研究。
★ 0 📥 579
dev-programming

Mcporter

steipete
使用 mcporter CLI 直接列出、配置、认证及调用 MCP 服务器/工具(支持 HTTP 或 stdio),涵盖临时服务器、配置编辑及 CLI/类型生成功能。
★ 199 📥 68,371
dev-programming

Github

steipete
使用 `gh` CLI 与 GitHub 交互,通过 `gh issue`、`gh pr`、`gh run` 和 `gh api` 管理议题、PR、CI 运行及高级查询。
★ 688 📥 331,815