How melihsenlen/sentinel_curve works
Sentinel Curve collects Windows CPU and memory measurements, trains an LSTM on the resulting time series, and produces autoregressive forecasts for visualization. The graph follows the documented monitor-to-CSV and train-to-inference workflows; source excerpts show the Python calls and file operations. The notebook’s visualization details and configuration values are not sampled, so those links rely on the README.
Main entry points: inference.py Autoregressive inferencer · train.py Trainer · main.cpp Monitor loop · monitor.cpp Resource sampler
Components 16
Resource monitoring
- Monitor loop
monitor/src/main.cpp - Resource sampler
monitor/src/monitor.cpp - Sample record
monitor/include/monitor.hpp - Measurements CSV
Model training
- Train and infer entry
pipeline/run.py - Trainer
architecture/train.py - Time-series reader
architecture/data.py - LSTM regressor
architecture/model.py - Configuration loader
architecture/config.py - Run configuration
config.yaml
Forecasting
- Trained weights
- Autoregressive inferencer
architecture/inference.py
Results and visualization
- Forecast CSV
- Trend visualization
analysis.ipynb
Other components
- Operator
- Windows resources
Connections 21
- Operator → Monitor loop: runs monitor
README.md - Monitor loop → Resource sampler: collects sample
monitor/src/main.cpp - Resource sampler → Windows resources: reads usage
monitor/src/monitor.cpp - Resource sampler → Sample record: returns sample
monitor/src/monitor.cpp - Monitor loop → Measurements CSV: writes CSV
monitor/src/main.cpp - Operator → Train and infer entry: runs pipeline
README.md - Train and infer entry → Trainer: runs training
pipeline/run.py - Train and infer entry → Autoregressive inferencer: runs inference
pipeline/run.py - Trainer → Configuration loader: loads settings
architecture/train.py - Configuration loader → Run configuration: reads YAML
architecture/config.py - Trainer → Time-series reader: builds sequences
architecture/train.py - Time-series reader → Measurements CSV: reads measurements
README.md - Trainer → LSTM regressor: trains model
architecture/train.py - Trainer → Trained weights: saves weights
architecture/train.py - Autoregressive inferencer → Configuration loader: loads settings
architecture/inference.py - Autoregressive inferencer → Time-series reader: creates sequences
architecture/inference.py - Autoregressive inferencer → LSTM regressor: predicts steps
architecture/inference.py - Autoregressive inferencer → Trained weights: loads weights
architecture/inference.py - Autoregressive inferencer → Forecast CSV: writes forecasts
architecture/inference.py - Operator → Trend visualization: views results
README.md - Trend visualization → Forecast CSV: visualizes forecasts
README.md