Logo

Machine Learning

Experiment Tracking

AI Generated Summary

Run #142 (gradient boosting, depth=8) is the best performer this week at 95.4% validation accuracy, a 1.2pt improvement over the current production model. 2 runs failed due to feature mismatch and have been flagged for review.

Confidence 93.8% · generated 7 min ago

Running

4

Completed Today

18

Failed

2

Reproducibility Score

96%

Experiment Comparison

Validation accuracy across recent runs

Metrics Comparison

Accuracy vs. precision vs. recall · top 3 runs

Hyperparameter Tracking

Learning rate vs. accuracy across all runs

Experiment Timeline

Run duration · last 24 hours

Dataset Versions

Rows1.2M
Features42
3 versions tracked Active: churn_q3_v2

Best Performing Runs

Run #142 · gradient boosting95.4%
Run #138 · random forest94.1%
Run #135 · XGBoost93.6%
Best vs. current production +1.2 pt

Failed Experiments

Run #140 · feature schema mismatch
Run #136 · out-of-memory on GPU
Run #129 · learning rate divergence
Failed this week 3 of 20 runs

AI Experiment Recommendations

Try depth=10 with L2 regularization — similar configs improved accuracy 0.8pt in comparable experiments.
Re-run Run #140 against churn_q3_v2 after the feature schema fix to confirm the mismatch is resolved.
2 recommendations available

Artifact Storage

Model Checkpoints

842 GB

Logs & Metrics

62 GB

Datasets Cached

1.2 TB

Retention Window

90 days

Total artifact storage used 2.1 TB

Experiment Notebook · Run #142

model = GradientBoostingClassifier(
    n_estimators=400, max_depth=8, learning_rate=0.05
)
model.fit(X_train, y_train)
# val_accuracy: 0.954

Experiment History

Run #142 completed · 95.4% val accuracy1 hour ago
Run #141 completed · 92.8% val accuracy3 hours ago
Run #140 failed · feature schema mismatch5 hours ago
Run #138 completed · 94.1% val accuracyYesterday
18 runs completed today