| license: mit | |
| library_name: xgboost | |
| tags: | |
| - sirna | |
| - rna | |
| - regression | |
| - xgboost | |
| - lightgbm | |
| # siRBench Model | |
| Runtime artifacts for the siRBench predictor Space. | |
| This repo is intended to store the files required for inference: | |
| - `xgb_model.json` | |
| - `lgbm_model.txt` | |
| - `calibrator.joblib` | |
| - `feature_artifacts.json` | |
| ## Intended consumer | |
| The main consumer is the Docker Space: | |
| - `dimostzim/siRBench-predictor` | |
| That app expects two input sequences: | |
| - `siRNA` | |
| - `mRNA` target window | |
| The app then computes the engineered feature set and loads these model artifacts | |
| from the Hub. | |
| ## Notes | |
| - This repo stores inference artifacts, not training code. | |
| - The final prediction is a calibrated ensemble of XGBoost and LightGBM models. | |