AGENT SKILL
explainability
AI model explainability
- License 许可
- MIT
- 最近更新
- 2026-03-01
- Stars
- ★ 3
- Forks
- 2
01 前言 Front Matter
- name
- explainability
- license
- MIT
- metadata
- { "audience": "machine-learning-engineers", "category": "artificial-intelligence" }
- description
- AI model explainability
- compatibility
- opencode
02 资源文件 2
- 脚本scripts/gtex_eqtl.py4.3 KB
- 脚本scripts/variant_resolution.py11.9 KB
03 SKILL.md 正文
What I do
- Explain model predictions
- Implement interpretable models
- Create feature importance analysis
- Build visualization tools
- Design trust-building explanations
- Handle regulatory requirements
When to use me
Use me when:
- Model debugging
- Stakeholder communication
- Regulatory compliance
- Bias detection
- Trust in AI decisions
Key Concepts
Explanation Methods
import shap
import lime
from sklearn.inspection import permutation_importance
# SHAP values
explainer = shap.TreeExplainer(model)
shap_values = explainer.shap_values(X_test)
# Visualize
shap.summary_plot(shap_values, X_test, feature_names=feature_names)
# LIME explanations
from lime.lime_tabular import LimeTabularExplainer
explainer = LimeTabularExplainer(X_train, feature_names=feature_names)
explanation = explainer.explain_instance(X_test[0], model.predict_proba)
# Permutation importance
result = permutation_importance(model, X_test, y_test, n_repeats=10)
Interpretable Models
- Linear models: Coefficients
- Decision trees: Path visualization
- Rule-based: Explicit rules
- Attention: Attention weights
Types of Explanations
- Global: Overall model behavior
- Local: Single prediction
- Feature: Feature importance
- Counterfactual: What if scenarios
04 同仓库更多 Skills
xamarin
Xamarin cross-platform development
★ 3
wrangling
Data wrangling fundamentals
★ 3
windows
Windows operating system administration and scripting
★ 3
wide-column
Wide-column data model
★ 3
wide-column-stores
Wide-column database stores
★ 3
websockets
WebSocket implementation and best practices
★ 3