How microsoft/mira works
MIRA is a medical time-series foundation model whose main workflow loads clinical sequences and timestamps, trains a time-aware model, and uses autoregressive or continuous-time forecasting for zero-shot evaluation. The repository also supports several dataset encodings and windowed/time-aware datasets. The graph preserves the documented CT-RoPE, frequency-specialized MoE, and neural ODE capabilities. Some dataset and evaluation modules were not excerpted, so their internal calls are not inferred.
Main entry points: modeling_mira.py MIRA model · mira_dataset.py Dataset aggregator · general_dataset.py File datasets · binary_dataset.py Binary dataset
Components 19
Clinical data
- Dataset aggregator
MIRA/datasets/mira_dataset.py - File datasets
MIRA/datasets/general_dataset.py - Binary dataset
MIRA/datasets/binary_dataset.py - Dataset interface
MIRA/datasets/ts_dataset.py - Time-aware data
MIRA/datasets/timeawared_dataset.py - Window datasets
MIRA/datasets/mira_window_dataset.py - Benchmark data
MIRA/datasets/benchmark_dataset.py - Time utilities
MIRA/datasets/time_utils.py
Training
- Training CLI
main.py - Training runner
MIRA/runner.py - HF trainer
MIRA/trainer/hf_trainer.py
Forecast model
- Model configuration
MIRA/models/configuration_mira.py - MIRA model
MIRA/models/modeling_mira.py - Generation mixin
MIRA/models/ts_generation_mixin.py - Time normalization
MIRA/models/utils_time_normalization.py
Inference and evaluation
- Autoregressive inference
MIRA/models/mira_inference.py - Evaluation CLI
model_eval.py - Evaluation runner
run_eval.py
Other components
- Researcher
Connections 14
- Researcher → Training CLI: starts training
README.md - Training CLI → Training runner: configures
main.py - Training runner → Time-aware data: loads time-aware data
MIRA/runner.py - Dataset aggregator → Binary dataset: selects
MIRA/datasets/mira_dataset.py - Dataset aggregator → File datasets: selects
MIRA/datasets/mira_dataset.py - Dataset aggregator → Time-aware data: loads JSONL
MIRA/datasets/mira_dataset.py - Binary dataset → Dataset interface: implements
MIRA/datasets/binary_dataset.py - File datasets → Dataset interface: implements
MIRA/datasets/general_dataset.py - Model configuration → MIRA model: configures
MIRA/models/modeling_mira.py - MIRA model → Generation mixin: uses
MIRA/models/modeling_mira.py - MIRA model → Time normalization: uses
MIRA/models/modeling_mira.py - Researcher → Autoregressive inference: runs forecasts
README.md - Autoregressive inference → MIRA model: calls
MIRA/models/mira_inference.py - Researcher → Evaluation CLI: runs evaluation
README.md