dimostzim/siRBench-data

--- configs: - config_name: default data_files: - split: train path: siRBench_train.csv - split: test path: siRBench_test.csv - split: leftout path: siRBench_leftout.csv --- # siRBench data This repository contains the three public siRBench tabular splits used by the model and benchmark workflows. Splits: - `train`: 2,776 rows - `test`: 275 rows - `leftout`: 896 rows All three files share the same 106-column schema, including: - `siRNA` - `mRNA` - `extended_mRNA` - `efficiency` - `source` - `cell_line` - engineered thermodynamic features `leftout` is a second held-out test set containing cell lines that do not appear in the train or test splits.